IEEE News

by Ian McKay on 24. June 2024. at 13:00

Imagine it’s 2050 and you’re on a cross-country flight on a new type of airliner, one with no fuel on board. The plane takes off, and you rise above the airport. Instead of climbing to cruising altitude, though, your plane levels out and the engines quiet to a low hum. Is this normal? No one seems to know. Anxious passengers crane their necks to get a better view out their windows. They’re all looking for one thing.

Then it appears: a massive antenna array on the horizon. It’s sending out a powerful beam of electromagnetic radiation pointed at the underside of the plane. After soaking in that energy, the engines power up, and the aircraft continues its climb. Over several minutes, the beam will deliver just enough energy to get you to the next ground antenna located another couple hundred kilometers ahead.

The person next to you audibly exhales. You sit back in your seat and wait for your drink. Old-school EV-range anxiety is nothing next to this.

Electromagnetic waves on the fly

Beamed power for aviation is, I admit, an outrageous notion. If physics doesn’t forbid it, federal regulators or nervous passengers probably will. But compared with other proposals for decarbonizing aviation, is it that crazy?

Batteries, hydrogen, alternative carbon-based fuels—nothing developed so far can store energy as cheaply and densely as fossil fuels, or fully meet the needs of commercial air travel as we know it. So, what if we forgo storing all the energy on board and instead beam it from the ground? Let me sketch what it would take to make this idea fly.

Beamed Power for Aviation

Fly by Microwave: Warm up to a new kind of air travel

For the wireless-power source, engineers would likely choose microwaves because this type of electromagnetic radiation can pass unruffled through clouds and because receivers on planes could absorb it completely, with nearly zero risk to passengers.

To power a moving aircraft, microwave radiation would need to be sent in a tight, steerable beam. This can be done using technology known as a phased array, which is commonly used to direct radar beams. With enough elements spread out sufficiently and all working together, phased arrays can also be configured to focus power on a point a certain distance away, such as the receiving antenna on a plane.

Phased arrays work on the principle of constructive and destructive interference. The radiation from the antenna elements will, of course, overlap. In some directions the radiated waves will interfere destructively and cancel out one another, and in other directions the waves will fall perfectly in phase, adding together constructively. Where the waves overlap constructively, energy radiates in that direction, creating a beam of power that can be steered electronically.

How far we can send energy in a tight beam with a phased array is governed by physics—specifically, by something called the diffraction limit. There’s a simple way to calculate the optimal case for beamed power: D1 D2 > λ R. In this mathematical inequality, D1 and D2 are the diameters of the sending and receiving antennas, λ is the wavelength of the radiation, and R is the distance between those antennas.

Now, let me offer some ballpark numbers to figure out how big the transmitting antenna (D1) must be. The size of the receiving antenna on the aircraft is probably the biggest limiting factor. A medium-size airliner has a wing and body area of about 1,000 square meters, which should provide for the equivalent of a receiving antenna that’s 30 meters wide (D2) built into the underside of the plane.

If physics doesn’t forbid it, federal regulators or nervous passengers probably will.

Next, let’s guess how far we would need to beam the energy. The line of sight to the horizon for someone in an airliner at cruising altitude is about 360 kilometers long, assuming the terrain below is level. But mountains would interfere, plus nobody wants range anxiety, so let’s place our ground antennas every 200 km along the flight path, each beaming energy half of that distance. That is, set R to 100 km.

Finally, assume the microwave wavelength (λ) is 5 centimeters. This provides a happy medium between a wavelength that’s too small to penetrate clouds and one that’s too large to gather back together on a receiving dish. Plugging these numbers into the equation above shows that in this scenario the diameter of the ground antennas (D1) would need to be at least about 170 meters. That’s gigantic, but perhaps not unreasonable. Imagine a series of three or four of these antennas, each the size of a football stadium, spread along the route, say, between LAX and SFO or between AMS and BER.

Power beaming in the real world

While what I’ve described is theoretically possible, in practice engineers have beamed only a fraction of the amount of power needed for an airliner, and they’ve done that only over much shorter distances.

NASA holds the record from an experiment in 1975, when it beamed 30 kilowatts of power over 1.5 km with a dish the size of a house. To achieve this feat, the team used an analog device called a klystron. The geometry of a klystron causes electrons to oscillate in a way that amplifies microwaves of a particular frequency—kind of like how the geometry of a whistle causes air to oscillate and produce a particular pitch.

Klystrons and their cousins, cavity magnetrons (found in ordinary microwave ovens), are quite efficient because of their simplicity. But their properties depend on their precise geometry, so it’s challenging to coordinate many such devices to focus energy into a tight beam.

In more recent years, advances in semiconductor technology have allowed a single oscillator to drive a large number of solid-state amplifiers in near-perfect phase coordination. This has allowed microwaves to be focused much more tightly than was possible before, enabling more-precise energy transfer over longer distances.

In 2022, the Auckland-based startup Emrod showed just how promising this semiconductor-enabled approach could be. Inside a cavernous hangar in Germany owned by Airbus, the researchers beamed 550 watts across 36 meters and kept over 95 percent of the energy flowing in a tight beam—far better than could be achieved with analog systems. In 2021, the U.S. Naval Research Laboratory showed that these techniques could handle higher power levels when it sent more than a kilowatt between two ground antennas over a kilometer apart. Other researchers have energized drones in the air, and a few groups even intend to use phased arrays to beam solar power from satellites to Earth.

A rectenna for the ages

So beaming energy to airliners might not be entirely crazy. But please remain seated with your seat belts fastened; there’s some turbulence ahead for this idea. A Boeing 737 aircraft at takeoff requires about 30 megawatts—a thousand times as much power as any power-beaming experiment has demonstrated. Scaling up to this level while keeping our airplanes aerodynamic (and flyable) won’t be easy.

Consider the design of the antenna on the plane, which receives and converts the microwaves to an electric current to power the aircraft. This rectifying antenna, or rectenna, would need to be built onto the underside surfaces of the aircraft with aerodynamics in mind. Power transmission will be maximized when the plane is right above the ground station, but it would be far more limited the rest of the time, when ground stations are far ahead or behind the plane. At those angles, the beam would activate only either the front or rear surfaces of the aircraft, making it especially hard to receive enough power.

With 30 MW blasting onto that small of an area, power density will be an issue. If the aircraft is the size of Boeing 737, the rectenna would have to cram about 25 W into each square centimeter. Because the solid-state elements of the array would be spaced about a half-wavelength—or 2.5 cm—apart, this translates to about 150 W per element—perilously close to the maximum power density of any solid-state power-conversion device. The top mark in the 2016 IEEE/Google Little Box Challenge was about 150 W per cubic inch (less than 10 W per cubic centimeter).

The rectenna will also have to weigh very little and minimize the disturbance to the airflow over the plane. Compromising the geometry of the rectenna for aerodynamic reasons might lower its efficiency. State-of-the art power-transfer efficiencies are only about 30 percent, so the rectenna can’t afford to compromise too much.

A Boeing 737 aircraft at takeoff requires about 30 megawatts—a thousand times as much power as any power-beaming experiment has demonstrated.

And all of this equipment will have to work in an electric field of about 7,000 volts per meter—the strength of the power beam. The electric field inside a microwave oven, which is only about a third as strong, can create a corona discharge, or electric arc, between the tines of a metal fork, so just imagine what might happen inside the electronics of the rectenna.

And speaking of microwave ovens, I should mention that, to keep passengers from cooking in their seats, the windows on any beamed-power airplane would surely need the same wire mesh that’s on the doors of microwave ovens—to keep those sizzling fields outside the plane. Birds, however, won’t have that protection.

Fowl flying through our power beam near the ground might encounter a heating of more than 1,000 watts per square meter—stronger than the sun on a hot day. Up higher, the beam will narrow to a focal point with much more heat. But because that focal point would be moving awfully fast and located higher than birds typically fly, any roasted ducks falling from the sky would be rare in both senses of the word. Ray Simpkin, chief science officer at Emrod, told me it’d take “more than 10 minutes to cook a bird” with Emrod’s relatively low-power system.

Legal challenges would surely come, though, and not just from the National Audubon Society. Thirty megawatts beamed through the air would be about 10 billion times as strong as typical signals at 5-cm wavelengths (a band currently reserved for amateur radio and satellite communications). Even if the transmitter could successfully put 99 percent of the waves into a tight beam, the 1 percent that’s leaked would still be a hundred million times as strong as approved transmissions today.

And remember that aviation regulators make us turn off our cellphones during takeoff to quiet radio noise, so imagine what they’ll say about subjecting an entire plane to electromagnetic radiation that’s substantially stronger than that of a microwave oven. All these problems are surmountable, perhaps, but only with some very good engineers (and lawyers).

Compared with the legal obstacles and the engineering hurdles we’d need to overcome in the air, the challenges of building transmitting arrays on the ground, huge as they would have to be, seem modest. The rub is the staggering number of them that would have to be built. Many flights occur over mountainous terrain, producing a line of sight to the horizon that is less than 100 km. So in real-world terrain we’d need more closely spaced transmitters. And for the one-third of airline miles that occur over oceans, we would presumably have to build floating arrays. Clearly, building out the infrastructure would be an undertaking on the scale of the Eisenhower-era U.S. interstate highway system.

Decarbonizing with the world’s largest microwave

People might be able to find workarounds for many of these issues. If the rectenna is too hard to engineer, for example, perhaps designers will find that they don’t have to turn the microwaves back into electricity—there are precedents for using heat to propel airplanes. A sawtooth flight path—with the plane climbing up as it approaches each emitter station and gliding down after it passes by—could help with the power-density and field-of-view issues, as could flying-wing designs, which have much more room for large rectennas. Perhaps using existing municipal airports or putting ground antennas near solar farms could reduce some of the infrastructure cost. And perhaps researchers will find shortcuts to radically streamline phased-array transmitters. Perhaps, perhaps.

To be sure, beamed power for aviation faces many challenges. But less-fanciful options for decarbonizing aviation have their own problems. Battery-powered planes don’t even come close to meeting the needs of commercial airlines. The best rechargeable batteries have about 5 percent of the effective energy density of jet fuel. At that figure, an all-electric airliner would have to fill its entire fuselage with batteries—no room for passengers, sorry—and it’d still barely make it a tenth as far as an ordinary jet. Given that the best batteries have improved by only threefold in the past three decades, it’s safe to say that batteries won’t power commercial air travel as we know it anytime soon.

Any roasted ducks falling from the sky would be rare in both senses of the word.

Hydrogen isn’t much further along, despite early hydrogen-powered flights occurring nearly 40 years ago. And it’s potentially dangerous—enough that some designs for hydrogen planes have included two separate fuselages: one for fuel and one for people to give them more time to get away if the stuff gets explode-y. The same factors that have kept hydrogen cars off the road will probably keep hydrogen planes out of the sky.

Synthetic and biobased jet fuels are probably the most reasonable proposal. They’ll give us aviation just as we know it today, just at a higher cost—perhaps 20 to 50 percent more expensive per ticket. But fuels produced from food crops can be worse for the environment than the fossil fuels they replace, and fuels produced from CO₂ and electricity are even less economical. Plus, all combustion fuels could still contribute to contrail formation, which makes up more than half of aviation’s climate impact.

The big problem with the “sane” approach for decarbonizing aviation is that it doesn’t present us with a vision of the future at all. At the very best, we’ll get a more expensive version of the same air travel experience the world has had since the 1970s.

True, beamed power is far less likely to work. But it’s good to examine crazy stuff like this from time to time. Airplanes themselves were a crazy idea when they were first proposed. If we want to clean up the environment and produce a future that actually looks like a future, we might have to take fliers on some unlikely sounding schemes.

by Willie D. Jones on 23. June 2024. at 18:00

Tsunenobu Kimoto, a professor of electronic science and engineering at Kyoto University, literally wrote the book on silicon carbide technology. Fundamentals of Silicon Carbide Technology, published in 2014, covers properties of SiC materials, processing technology, theory, and analysis of practical devices.

Kimoto, whose silicon carbide research has led to better fabrication techniques, improved the quality of wafers and reduced their defects. His innovations, which made silicon carbide semiconductor devices more efficient and more reliable and thus helped make them commercially viable, have had a significant impact on modern technology.

Tsunenobu Kimoto

Employer

Kyoto University

Title

Professor of electronic science and engineering

Member grade

Fellow

Alma mater

Kyoto University

For his contributions to silicon carbide material and power devices, the IEEE Fellow was honored with this year’s IEEE Andrew S. Grove Award, sponsored by the IEEE Electron Devices Society.

Silicon carbide’s humble beginnings

Decades before a Tesla Model 3 rolled off the assembly line with an SiC inverter, a small cadre of researchers, including Kimoto, foresaw the promise of silicon carbide technology. In obscurity they studied it and refined the techniques for fabricating power transistors with characteristics superior to those of the silicon devices then in mainstream use.

Today MOSFETs and other silicon carbide transistors greatly reduce on-state loss and switching losses in power-conversion systems, such as the inverters in an electric vehicle used to convert the battery’s direct current to the alternating current that drives the motor. Lower switching losses make the vehicles more efficient, reducing the size and weight of their power electronics and improving power-train performance. Silicon carbide–based chargers, which convert alternating current to direct current, provide similar improvements in efficiency.

But those tools didn’t just appear. “We had to first develop basic techniques such as how to dope the material to make n-type and p-type semiconductor crystals,” Kimoto says. N-type crystals’ atomic structures are arranged so that electrons, with their negative charges, move freely through the material’s lattice. Conversely, the atomic arrangement of p-type crystals’ contains positively charged holes.

Kimoto’s interest in silicon carbide began when he was working on his Ph.D. at Kyoto University in 1990.

“At that time, few people were working on silicon carbide devices,” he says. “And for those who were, the main target for silicon carbide was blue LED.

“There was hardly any interest in silicon carbide power devices, like MOSFETs and Schottky barrier diodes.”

Kimoto began by studying how SiC might be used as the basis of a blue LED. But then he read B. Jayant Baliga’s 1989 paper “Power Semiconductor Device Figure of Merit for High-Frequency Applications” in IEEE Electron Device Letters, and he attended a presentation by Baliga, the 2014 IEEE Medal of Honor recipient, on the topic.

“I was convinced that silicon carbide was very promising for power devices,” Kimoto says. “The problem was that we had no wafers and no substrate material,” without which it was impossible to fabricate the devices commercially.

In order to get silicon carbide power devices, “researchers like myself had to develop basic technology such as how to dope the material to make p-type and n-type crystals,” he says. “There was also the matter of forming high-quality oxides on silicon carbide.” Silicon dioxide is used in a MOSFET to isolate the gate and prevent electrons from flowing into it.

The first challenge Kimoto tackled was producing pure silicon carbide crystals. He decided to start with carborundum, a form of silicon carbide commonly used as an abrasive. Kimoto took some factory waste materials—small crystals of silicon carbide measuring roughly 5 millimeters by 8 mm­—and polished them.

He found he had highly doped n-type crystals. But he realized having only highly doped n-type SiC would be of little use in power applications unless he also could produce lightly doped (high purity) n-type and p-type SiC.

Connecting the two material types creates a depletion region straddling the junction where the n-type and p-type sides meet. In this region, the free, mobile charges are lost because of diffusion and recombination with their opposite charges, and an electric field is established that can be exploited to control the flow of charges across the boundary.

“Silicon carbide is a family with many, many brothers.”

By using an established technique, chemical vapor deposition, Kimoto was able to grow high-purity silicon carbide. The technique grows SiC as a layer on a substrate by introducing gasses into a reaction chamber.

At the time, silicon carbide, gallium nitride, and zinc selenide were all contenders in the race to produce a practical blue LED. Silicon carbide, Kimoto says, had only one advantage: It was relatively easy to make a silicon carbide p-n junction. Creating p-n junctions was still difficult to do with the other two options.

By the early 1990s, it was starting to become clear that SiC wasn’t going to win the blue-LED sweepstakes, however. The inescapable reality of the laws of physics trumped the SiC researchers’ belief that they could somehow overcome the material’s inherent properties. SiC has what is known as an indirect band gap structure, so when charge carriers are injected, the probability of the charges recombining and emitting photons is low, leading to poor efficiency as a light source.

While the blue-LED quest was making headlines, many low-profile advances were being made using SiC for power devices. By 1993, a team led by Kimoto and Hiroyuki Matsunami demonstrated the first 1,100-volt silicon carbide Schottky diodes, which they described in a paper in IEEE Electron Device Letters. The diodes produced by the team and others yielded fast switching that was not possible with silicon diodes.

“With silicon p-n diodes,” Kimoto says, “we need about a half microsecond for switching. But with a silicon carbide, it takes only 10 nanoseconds.”

The ability to switch devices on and off rapidly makes power supplies and inverters more efficient because they waste less energy as heat. Higher efficiency and less heat also permit designs that are smaller and lighter. That’s a big deal for electric vehicles, where less weight means less energy consumption.

Kimoto’s second breakthrough was identifying which form of the silicon carbide material would be most useful for electronics applications.

“Silicon carbide is a family with many, many brothers,” Kimoto says, noting that more than 100 variants with different silicon-carbon atomic structures exist.

The 6H-type silicon carbide was the default standard phase used by researchers targeting blue LEDs, but Kimoto discovered that the 4H-type has much better properties for power devices, including high electron mobility. Now all silicon carbide power devices and wafer products are made with the 4H-type.

Silicon carbide power devices in electric vehicles can improve energy efficiency by about 10 percent compared with silicon, Kimoto says. In electric trains, he says, the power required to propel the cars can be cut by 30 percent compared with those using silicon-based power devices.

Challenges remain, he acknowledges. Although silicon carbide power transistors are used in Teslas, other EVs, and electric trains, their performance is still far from ideal because of defects present at the silicon dioxide–SiC interface, he says. The interface defects lower the performance and reliability of MOS-based transistors, so Kimoto and others are working to reduce the defects.

A career sparked by semiconductors

When Kimoto was an only child growing up in Wakayama, Japan, near Osaka, his parents insisted he study medicine, and they expected him to live with them as an adult. His father was a garment factory worker; his mother was a homemaker. His move to Kyoto to study engineering “disappointed them on both counts,” he says.

His interest in engineering was sparked, he recalls, when he was in junior high school, and Japan and the United States were competing for semiconductor industry supremacy.

At Kyoto University, he earned bachelor’s and master’s degrees in electrical engineering, in 1986 and 1988. After graduating, he took a job at Sumitomo Electric Industries’ R&D center in Itami. He worked with silicon-based materials there but wasn’t satisfied with the center’s research opportunities.

He returned to Kyoto University in 1990 to pursue his doctorate. While studying power electronics and high-temperature devices, he also gained an understanding of material defects, breakdown, mobility, and luminescence.

“My experience working at the company was very valuable, but I didn’t want to go back to industry again,” he says. By the time he earned his doctorate in 1996, the university had hired him as a research associate.

He has been there ever since, turning out innovations that have helped make silicon carbide an indispensable part of modern life.

Growing the silicon carbide community at IEEE

Kimoto joined IEEE in the late 1990s. An active volunteer, he has helped grow the worldwide silicon carbide community.

He is an editor of IEEE Transactions on Electron Devices, and he has served on program committees for conferences including the International Symposium on Power Semiconductor Devices and ICs and the IEEE Workshop on Wide Bandgap Power Devices and Applications.

“Now when we hold a silicon carbide conference, more than 1,000 people gather,” he says. “At IEEE conferences like the International Electron Devices Meeting or ISPSD, we always see several well-attended sessions on silicon carbide power devices because more IEEE members pay attention to this field now.”

by Evan Ackerman on 21. June 2024. at 18:33

Video Friday is your weekly selection of awesome robotics videos, collected by your friends at IEEE Spectrum robotics. We also post a weekly calendar of upcoming robotics events for the next few months. Please send us your events for inclusion.

RoboCup 2024: 17–22 July 2024, EINDHOVEN, NETHERLANDS

ICRA@40: 23–26 September 2024, ROTTERDAM, NETHERLANDS

IROS 2024: 14–18 October 2024, ABU DHABI, UAE

ICSR 2024: 23–26 October 2024, ODENSE, DENMARK

Cybathlon 2024: 25–27 October 2024, ZURICH

Enjoy today’s videos!

We present Morphy, a novel compliant and morphologically aware flying robot that integrates sensorized flexible joints in its arms, thus enabling resilient collisions at high speeds and the ability to squeeze through openings narrower than its nominal dimensions.

Morphy represents a new class of soft flying robots that can facilitate unprecedented resilience through innovations both in the “body” and “brain.” The novel soft body can, in turn, enable new avenues for autonomy. Collisions that previously had to be avoided have now become acceptable risks, while areas that are untraversable for a certain robot size can now be negotiated through self-squeezing. These novel bodily interactions with the environment can give rise to new types of embodied intelligence.

[ ARL ]

Thanks, Kostas!

Segments of daily training for robots driven by reinforcement learning. Multiple tests done in advance for friendly service humans. The training includes some extreme tests. Please do not imitate!

[ Unitree ]

Sphero is not only still around, it’s making new STEM robots!

[ Sphero ]

Googly eyes mitigate all robot failures.

[ WVUIRL ]

Here I am, without the ability or equipment (or desire) required to iron anything that I own, and Flexiv’s got robots out there ironing fancy leather car seats.

[ Flexiv ]

Thanks, Noah!

We unveiled a significant leap forward in perception technology for our humanoid robot GR-1. The newly adapted pure-vision solution integrates bird’s-eye view, transformer models, and an occupancy network for precise and efficient environmental perception.

[ Fourier ]

Thanks, Serin!

LimX Dynamics’ humanoid robot CL-1 was launched in December 2023. It climbed stairs based on real-time terrain perception, two steps per stair. Four months later, in April 2024, the second demo video showcased CL-1 in the same scenario. It had advanced to climb the same stair, one step per stair.

[ LimX Dynamics ]

Thanks, Ou Yan!

New research from the University of Massachusetts Amherst shows that programming robots to create their own teams and voluntarily wait for their teammates results in faster task completion, with the potential to improve manufacturing, agriculture, and warehouse automation.

[ HCRL ] via [ UMass Amherst ]

Thanks, Julia!

LASDRA (Large-size Aerial Skeleton with Distributed Rotor Actuation system (ICRA18) is a scalable and modular aerial robot. It can assume a very slender, long, and dexterous form factor and is very lightweight.

[ SNU INRoL ]

We propose augmenting initially passive structures built from simple repeated cells, with novel active units to enable dynamic, shape-changing, and robotic applications. Inspired by metamaterials that can employ mechanisms, we build a framework that allows users to configure cells of this passive structure to allow it to perform complex tasks.

[ CMU ]

Testing autonomous exploration at the Exyn Office using Spot from Boston Dynamics. In this demo, Spot autonomously explores our flight space while on the hunt for one of our engineers.

[ Exyn ]

Meet Heavy Picker, the strongest robot in bulky-waste sorting and an absolute pro at lifting and sorting waste. With skills that would make a concert pianist jealous and a work ethic that never needs coffee breaks, Heavy Picker was on the lookout for new challenges.

[ Zen Robotics ]

AI is the biggest and most consequential business, financial, legal, technological, and cultural story of our time. In this panel, you will hear from the underrepresented community of women scientists who have been leading the AI revolution—from the beginning to now.

[ Stanford HAI ]

by Joanna Goodrich on 20. June 2024. at 18:00

Lynn Conway, codeveloper of very-large-scale integration, died on 9 June at the age of 86. The VLSI process, which creates integrated circuits by combining thousands of transistors into a single chip, revolutionized microchip design.

Conway, an IEEE Fellow, was transfeminine and was a transgender-rights activist who played a key role in updating the IEEE Code of Conduct to prohibit discrimination based on sexual orientation, gender identity, and gender expression.

She shared her experiences on a blog to help others considering or beginning to transition their gender identity. She also mentored many trans people through their transitioning.

“Lynn Conway’s example of engineering impact and personal courage has been a great source of inspiration for me and countless others,” Michael Wellman, a professor of computer science and engineering at the University of Michigan in Ann Arbor, told the Michigan Engineering News website. Conway was a professor emerita at the university.

The profile of Conway below is based on an interview The Institute conducted with her in December.

Some engineers dream their pioneering technologies will one day earn them a spot in history books. But what happens when your contributions are overlooked because of your gender identity?

If you’re like Lynn Conway—who faced that dilemma—you fight back.

Conway helped develop very-large-scale integration: the process of creating integrated circuits by combining thousands of transistors into a single chip. VLSI chips are at the core of electronic devices used today. The technology provides processing power, memory, and other functionalities to smartphones, laptops, smartwatches, televisions, and household appliances.

She and her research partner Carver Mead developed VLSI in the 1970s while she was working at Xerox’s Palo Alto Research Center, in California. Mead was an engineering professor at CalTech at the time. For years, Conway’s role was overlooked partly because she was a woman, she asserts, and partly because she was transfeminine.

Since coming out publicly in 1999, Conway has been fighting for her contributions to be recognized, and she’s succeeding. Over the years, the IEEE Fellow has been honored by a variety of organizations, most recently the National Inventors Hall of Fame, which inducted her last year almost 15 years after it recognized Mead.

From budding physicist to electrical engineer

Conway initially was interested in studying physics because of the role it played in World War II.

“After the war ended, physicists became famous for blowing up the world in order to save it,” she says. “I was naive and saw physics as the source of all wisdom. I went off to MIT, not fully understanding the subject I chose to major in.”

She took many electrical engineering courses because, she says, they allowed her to be creative. It was through those classes that she found her calling.

She left MIT in 1957, then earned bachelor’s and master’s degrees in electrical engineering from Columbia in 1962 and 1963. While at Columbia, she conducted an independent study under the guidance of Herb Schorr, an adjunct professor and a researcher at IBM Research in Yorktown Heights, N.Y. The study involved installing a list-processing language on the IBM 1620 computer, “which was the most arcane machine to attempt to do that on,” she says laughing. “It was a cool language that Maurice Wilkes from Cambridge had developed to experiment with self-compiling compilers.”

She must have made quite an impression on Schorr, she says, because after she earned her master’s degree, he recruited her to join him at the research center. While working on the advanced computing systems project there, she invented multiple-out-of-order dynamic instruction scheduling, a technique that allows a CPU to reorder instructions based on their availability and readiness instead of following the program order strictly.

That work led to the creation of the superscalar CPU, which manages multiple instruction pipelines to execute several instructions concurrently.

The company eventually transferred her to its offices in California’s Bay Area.

Although her career was thriving, Conway was struggling with gender dysphoria, the distress people experience when their gender identity differs from their sex assigned at birth. In 1967 she moved forward with gender-affirming care “to resolve the terrible existential situation I had faced since childhood,” she says.

She notified IBM of her intention to transition, with the hope the company would allow her to do so quietly. Instead, IBM fired her, convinced that her transition would cause “extreme emotional distress in fellow employees,” she says. (In 2020 the company issued an apology for terminating her.)

After completing her transition, at the end of 1968 Conway began her career anew as a contract programmer. By 1971 she was working as a computer architect at Memorex in Silicon Valley. She joined the company in what she calls “stealth mode.” No one other than close family members and friends knew she was transfeminine. Conway was afraid of discrimination and losing her job again, she says. Because of her decision to keep her transition a secret, she says, she could not claim credit for the techniques she had invented at IBM Research because they were credited to the name she had been assigned at birth, her “dead name.”

She was recruited in 1975 to join Xerox PARC as a research fellow and manager of its VLSI system design group.

It was there that she made history.

Conway was recruited in 1975 to join Xerox PARC as a research fellow.Lynn Conway

Starting the Mead and Conway Revolution

Concerned with how Moore’s Law would affect the performance of microelectronics, the Advanced Research Project Agency (now known as the Defense Advanced Research Projects Agency) created a coalition of companies and research universities, including PARC and CalTech, to improve microchip design. After Conway joined PARC’s VLSI system design group, she worked closely with Carver Mead on chip design. Mead, now an IEEE Life Fellow, is credited with coining the term Moore’s Law.

Making chips at the time involved manually designing transistors and connecting them with circuits. The process was time-consuming and error-prone.

“A whole bunch of different pieces of design were being done at different abstraction levels, including the basic architecture, the logic design, the circuit design, and the layout design—all by different people,” Conway said in a 2023 IEEE Annals of the History of Computing interview. “And the various people in the different layers passed the design down in kind of a paternalistic top-down system. The people at any one layer may have no clue what the people at the other levels in that system are doing or what they know.”

Conway and Mead decided the best way to address that communication problem was to use CAD tools to automate the process.

The two also introduced the structured-design method of creating chips. It emphasized high-level abstraction and modular design techniques such as logic gates and modules—which made the process more efficient and scalable.

Conway also created a simplified set of rules for chip design that enabled the integrated circuits to be numerically encoded, scaled, and reused as Moore’s Law advanced.

The method was so radical, she says, that it needed help catching on. Conway and Mead wrote Introduction to VLSI Systems to take the new concepts straight to the next generation of engineers and programmers. The textbook included the basics of structured designs and how to validate and verify them. Before its publication in 1980, Conway tested how well it explained the method by teaching the first VLSI course in 1978 at MIT.

The textbook was successful, becoming the foundational resource for teaching the technology. By 1983 it was being used by nearly 120 universities.

Conway and Mead’s work resulted in what is known as the Mead and Conway Revolution, enabling faster, smaller, and more powerful devices to be developed.

Throughout the 1980s, Conway and Mead were known as the dynamic duo that created VLSI. They received multiple joint awards including the Electronics magazine 1981 Award for Achievement, the University of Pennsylvania’s 1984 Pender Award, and the Franklin Institute’s 1985 Wetherill Medal.

Conway left Xerox PARC in 1983 to join DARPA as assistant director for strategic computing. She led planning of the strategic computing initiative, an effort to expand the technology base for intelligent-weapons systems.

Two years later she began her academic career at the University of Michigan as a professor of electrical engineering and computer science. She was the university’s associate dean of engineering and taught there until 1998, when she retired.

Becoming an activist

In 1999 Conway decided to come out as a transfeminine engineer, knowing that not only would her previous work be credited to her again, she says, but also that she could be a source of strength and inspiration for others like her.

In the 2000s Conway’s honors began to dry up, while Mead continued to receive awards for VLSI, including a 2002 U.S. National Medal of Technology and Innovation.

After publicly coming out, she spoke openly about her experience and lobbied to be credited for her work.

Some organizations, including IEEE, began to recognize Conway. The IEEE Computer Society awarded her its 2009 Computer Pioneer Award. She received the 2015 IEEE/RSE Maxwell Medal, which honors contributions that had an exceptional impact on the development of electronics and electrical engineering.

by Shubham Agarwal on 20. June 2024. at 16:00

Tech companies have been caught up in a race to build the biggest large language models (LLMs). In April, for example, Meta announced the 400-billion-parameter Llama 3, which contains twice the number of parameters—or variables that determine how the model responds to queries—than OpenAI’s original ChatGPT model from 2022. Although not confirmed, GPT-4 is estimated to have about 1.8 trillion parameters.

In the last few months, however, some of the largest tech companies, including Apple and Microsoft, have introduced small language models (SLMs). These models are a fraction of the size of their LLM counterparts and yet, on many benchmarks, can match or even outperform them in text generation.

On 10 June, at Apple’s Worldwide Developers Conference, the company announced its “Apple Intelligence” models, which have around 3 billion parameters. And in late April, Microsoft released its Phi-3 family of SLMs, featuring models housing between 3.8 billion and 14 billion parameters.

OpenAI’s CEO Sam Altman believes we’re at the end of the era of giant models.

In a series of tests, the smallest of Microsoft’s models, Phi-3-mini, rivalled OpenAI’s GPT-3.5 (175 billion parameters), which powers the free version of ChatGPT, and outperformed Google’s Gemma (7 billion parameters). The tests evaluated how well a model understands language by prompting it with questions about mathematics, philosophy, law, and more. What’s more interesting, Microsoft’s Phi-3-small, with 7 billion parameters, fared remarkably better than GPT-3.5 in many of these benchmarks.

Aaron Mueller, who researches language models at Northeastern University in Boston, isn’t surprised SLMs can go toe-to-toe with LLMs in select functions. He says that’s because scaling the number of parameters isn’t the only way to improve a model’s performance: Training it on higher-quality data can yield similar results too.

Microsoft’s Phi models were trained on fine-tuned “textbook-quality” data, says Mueller, which have a more consistent style that’s easier to learn from than the highly diverse text from across the Internet that LLMs typically rely on. Similarly, Apple trained its SLMs exclusively on richer and more complex datasets.

The rise of SLMs comes at a time when the performance gap between LLMs is quickly narrowing and tech companies look to deviate from standard scaling laws and explore other avenues for performance upgrades. At an event in April, OpenAI’s CEO Sam Altman said he believes we’re at the end of the era of giant models. “We’ll make them better in other ways.”

Because SLMs don’t consume nearly as much energy as LLMs, they can also run locally on devices like smartphones and laptops (instead of in the cloud) to preserve data privacy and personalize them to each person. In March, Google rolled out Gemini Nano to the company’s Pixel line of smartphones. The SLM can summarize audio recordings and produce smart replies to conversations without an Internet connection. Apple is expected to follow suit later this year.

More importantly, SLMs can democratize access to language models, says Mueller. So far, AI development has been concentrated into the hands of a couple of large companies that can afford to deploy high-end infrastructure, while other, smaller operations and labs have been forced to license them for hefty fees.

Since SLMs can be easily trained on more affordable hardware, says Mueller, they’re more accessible to those with modest resources and yet still capable enough for specific applications.

In addition, while researchers agree there’s still a lot of work ahead to overcome hallucinations, carefully curated SLMs bring them a step closer toward building responsible AI that is also interpretable, which would potentially allow researchers to debug specific LLM issues and fix them at the source.

For researchers like Alex Warstadt, a computer science researcher at ETH Zurich, SLMs could also offer new, fascinating insights into a longstanding scientific question: How children acquire their first language. Warstadt, alongside a group of researchers including Northeastern’s Mueller, organizes BabyLM, a challenge in which participants optimize language-model training on small data.

Not only could SLMs potentially unlock new secrets of human cognition, but they also help improve generative AI. By the time children turn 13, they’re exposed to about 100 million words and are better than chatbots at language, with access to only 0.01 percent of the data. While no one knows what makes humans so much more efficient, says Warstadt, “reverse engineering efficient humanlike learning at small scales could lead to huge improvements when scaled up to LLM scales.”

by Margo Anderson on 20. June 2024. at 15:01

As the world’s 5G rollout continues with its predictable fits and starts, the cellular technology is also starting to move into a space already dominated by another wireless tech: Wi-Fi. Private 5G networks—in which a person or company sets up their own facility-wide cellular network—are today finding applications where Wi-Fi was once the only viable game in town. This month, the Newbury, England–based telecom company Vodafone is releasing a Raspberry Pi–based private 5G base station that it is now being aimed at developers, who might then jump-start a wave of private 5G innovation.

“The Raspberry Pi is the most affordable CPU[-based] computer that you can get,” says Santiago Tenorio, network architecture director at Vodafone. “Which means that what we build, in essence, has a similar bill of materials as a good quality Wi-Fi router.”

The company has teamed with the Surrey, England–based Lime Microsystems to release a crowd-funded range of private 5G base-station kits ranging in price from US $800 to $12,000.

“In a Raspberry Pi—in this case, a Raspberry Pi 4 is what we used—then you can be sure you can run that anywhere, because it’s the tiniest processor that you can have,” Tenorio says.

Santiago Tenorio holds one of Lime Microsystems’ private 5G base-station kits.Vodafone

Private 5G for Drones and Bakeries

There are a range of reasons, Tenorio says, why someone might want their own private 5G network. At the moment, the scenarios mostly concern companies and organizations—although individual expert users could still be drawn to, for instance, 5G’s relatively low latency and network flexibility.

Tenorio highlighted security and mobility as two big selling points for private 5G.

A commercial storefront business, for instance, might be attracted to the extra security protections that a SIM card can provide compared to password-based wireless network security. Because each SIM card contains its own unique identifier and encryption keys, thereby also enabling a network to be able to recognize and authorize each individual connection, Tenorio says private 5G network security is a considerable selling point.

Plus, Tenorio says, it’s simpler for customers to access. Envisioning a use case of a bakery with its own privately deployed 5G network, he says, “You don’t need a password. You don’t need a conversation [with a clerk behind a counter] or a QR code. You simply walk into the bakery, and you are into the bakery’s network.”

As to mobility, Tenorio suggests one emergency relief and rescue application that might rely on the presence of a nearby 5G station that causes devices in its range to ping.

Setting up a private 5G base station on a drone, Tenorio says, would enable that drone to fly over a disaster area and, via its airborne network, send a challenge signal to all devices in its coverage area to report in. Any device receiving that signal with a compatible SIM card then responds with its unique identification information.

“Then any phone would try to register,” Tenorio says. “And then you would know if there is someone [there].”

Not only that, but because the ping would be from a device with a SIM card, the private 5G rescue drone in the above scenario could potentially provide crucial information about each individual, just based on the device’s identifier alone. And that user-identifying feature of private 5G isn’t exactly irrelevant to a bakery owner—or to any other commercial customer—either, Tenorio says.

“If you are a bakery,” he says, “You could actually know who your customers are, because anyone walking into the bakery would register on your network and would leave their [International Mobile Subscriber Identity].”

Winning the Lag Race

According to Christian Wietfeld, professor of electrical engineering at the Technical University of Dortmund in Germany, private 5G networks also bring the possibility of less lag. His team has tested private 5G deployments—although, Wietfeld says that they haven’t yet tested the present Vodafone/Lime Microsystem base station—and have found private 5G to provide reliably better connectivity.

Wietfeld’s team will present their research at the IEEE International Symposium on Personal, Indoor and Mobile Radio Communications in September in Valencia, Spain. They found that private 5G can deliver connections up to 10 times as fast as connections in networks with high loads, compared to Wi-Fi (the IEEE 802.11 wireless standard).

“The additional cost and effort to operate a private 5G network pays off in lower downtimes of production and less delays in delivery of goods,” Wietfeld says. “Also, for safety-critical use cases such as on-campus teleoperated driving, private 5G networks provide the necessary reliability and predictability of performance.”

For Lime Networks, according to the company’s CEO and founder Ebrahim Bushehri, the challenge comes in developing a private 5G base station that maximized versatility and openness to whatever kinds of applications developers could envision—while still being reasonably inexpensive and retaining a low-power envelope.

“The solution had to be ultraportable and with an optional battery pack which could be mounted on drones and autonomous robots, for remote and tactical deployments, such as emergency-response scenarios and temporary events,” Bushehri says.

Meanwhile, the crowdfunding behind the device’s rollout, via the website Crowd Supply, allows both companies to keep tabs on the kinds of applications the developer community is envisioning for this technology, he says.

“Crowdfunding,” Bushehri says, “Is one of the key indicators of community interest and engagement. Hence the reason for launching the campaign on Crowd Supply to get feedback from early adopters.”

by Robert Schneider on 19. June 2024. at 18:00

When it comes to addressing climate change, the “in unity there’s strength” adage certainly applies.

To support IEEE’s climate change initiative, which highlights innovative solutions and approaches to the climate crisis, IEEE’s TryEngineering program has created a collection of lesson plans, activities, and events that cover electric vehicles, solar and wind power systems, and more.

TryEngineering, a program within IEEE Educational Activities, aims to foster the next generation of technology innovators by providing preuniversity educators and students with resources.

To help bring the climate collection to more students, TryEngineering has partnered with the Museum of Science in Boston. The museum, one of the world’s largest science centers, reaches nearly 5 million people annually through its physical location, nearby classrooms, and online platforms.

TryEngineering worked with the museum to distribute a nearly four-minute educational video created by Moment Factory, a multimedia studio specializing in immersive experiences. Using age-appropriate language, the video, which is posted on TryEngineering’s climate change page, explores the issue through visual models and scientific explanations.

“Since the industrial revolution, humans have been digging up fossil fuels and burning them, which releases CO2 into the atmosphere in unprecedented quantities,” the video says. It notes that in the past 60 years, atmospheric carbon dioxide increased at a rate 100 times faster than previous natural changes.

“We are committed to energizing students around important issues like climate change and helping them understand how engineering can make a difference.”

The video explains the impact of pollutants such as lead and ash, and it adds that “when we work together, we can change the global environment.” The video encourages students to contribute to a global solution by making small, personal changes.

“We’re thrilled to contribute to the IEEE climate change initiative by providing IEEE volunteers and educators access to TryEngineering’s collection, so they have resources to use with students,” says Debra Gulick, director of IEEE student and academic education programs.

“We are excited to partner with the Museum of Science to bring even more awareness and exposure of this important issue to the school setting,” Gulick says. “Working with prominent partners like the museum, we are committed to energizing students around important issues like climate change and helping them understand how engineering can make a difference.”

by Willie D. Jones on 19. June 2024. at 16:00

Earlier this month, China announced that it is pouring 6 billion yuan (about US $826 million) into a fund meant to spur the development of solid-state batteries by the nation’s leading battery manufacturers. Solid-state batteries use electrolytes of either glass, ceramic, or solid polymer material instead of the liquid lithium salts that are in the vast majority of today’s electric vehicle (EV) batteries. They’re greatly anticipated because they will have three or four times as much energy density as batteries with liquid electrolytes, offer more charge-discharge cycles over their lifetimes, and be far less susceptible to the thermal runaway reaction that occasionally causes lithium batteries to catch fire.

But China’s investment in the future of batteries won’t likely speed up the timetable for mass production and use in production vehicles. As IEEE Spectrum pointed out in January, it’s not realistic to look for solid-state batteries in production vehicles anytime soon. Experts Spectrum consulted at the time “noted a pointed skepticism toward the technical merits of these announcements. None could isolate anything on the horizon indicating that solid-state technology can escape the engineering and ‘production hell’ that lies ahead.”

“To state at this point that any one battery and any one country’s investments in battery R&D will dominate in the future is simply incorrect.” —Steve W. Martin, Iowa State University

Reaching scale production of solid-state batteries for EVs will first require validating existing solid-state battery technologies—now being used for other, less demanding applications—in terms of performance, life-span, and relative cost for vehicle propulsion. Researchers must still determine how those batteries take and hold a charge and deliver power as they age. They’ll also need to provide proof that a glass or ceramic battery can stand up to the jarring that comes with driving on bumpy roads and certify that it can withstand the occasional fender bender.

Here Come Semi-Solid-State Batteries

Meanwhile, as the world waits for solid electrolytes to shove liquids aside, Chinese EV manufacturer Nio and battery maker WeLion New Energy Technology Co. have partnered to stake a claim on the market for a third option that splits the difference: semi-solid-state batteries, with gel electrolytes.

CarNewsChina.com reported in April that the WeLion cells have an energy density of 360 watt-hours per kilogram. Fully packaged, the battery’s density rating is 260 Wh/kg. That’s still a significant improvement over lithium iron phosphate batteries, whose density tops out at 160 Wh/kg. In tests conducted last month with Nio’s EVs in Shanghai, Chengdu, and several other cities, the WeLion battery packs delivered more than 1,000 kilometers of driving range on a single charge. Nio says it plans to roll out the new battery type across its vehicle lineup beginning this month.

But the Beijing government’s largesse and the Nio-WeLion partnership’s attempt to be first to get semi-solid-state batteries into production vehicles shouldn’t be a temptation to call the EV propulsion game prematurely in China’s favor.

So says Steve W. Martin, a professor of materials science and engineering at Iowa State University, in Ames. Martin, whose research areas include glassy solid electrolytes for solid-state lithium batteries and high-capacity reversible anodes for lithium batteries, believes that solid-state batteries are the future and that hybrid semi-solid batteries will likely be a transition between liquid and solid-state batteries. However, he says, “to state at this point that any one battery and any one country’s investments in battery R&D will dominate in the future is simply incorrect.” Martin explains that “there are too many different kinds of solid-state batteries being developed right now and no one of these has a clear technological lead.”

The Advantages of Semi-Solid-State Batteries

The main innovation that gives semi-solid-state batteries an advantage over conventional batteries is the semisolid electrolyte from which they get their name. The gel electrolyte contains ionic conductors such as lithium salts just as liquid electrolytes do, but the way they are suspended in the gel matrix supports much more efficient ion conductivity. Enhanced transport of ions from one side of the battery to the other boosts the flow of current in the opposite direction that makes a complete circuit. This is important during the charging phase because the process happens more rapidly than it can in a battery with a liquid electrolyte. The gel’s structure also resists the formation of dendrites, the needlelike structures that can form on the anode during charging and cause short circuits. Additionally, gels are less volatile than liquid electrolytes and are therefore less prone to catching fire.

Though semi-solid-state batteries won’t reach the energy densities and life-spans that are expected from those with solid electrolytes, they’re at an advantage in the short term because they can be made on conventional lithium-ion battery production lines. Just as important, they have been tested and are available now rather than at some as yet unknown date.

Semi-solid-state batteries can be made on conventional lithium-ion battery production lines.

Several companies besides WeLion are actively developing semi-solid-state batteries. China’s prominent battery manufacturers, including CATL, BYD, and the state-owned automakers FAW Group and SAIC Group are, like WeLion, beneficiaries of Beijing’s plans to advance next-generation battery technology domestically. Separately, the startup Farasis Energy, founded in Ganzhou, China, in 2009, is collaborating with Mercedes-Benz to commercialize advanced batteries.

The Road Forward to Solid-State Batteries

U.S. startup QuantumScape says the solid-state lithium metal batteries it’s developing will offer energy density of around 400 Wh/kg. The company notes that its cells eliminate the charging bottleneck that occurs in conventional lithium-ion cells, where lithium must diffuse into the carbon particles. QuantumScape’s advanced batteries will therefore allow fast charging from 10 to 80 percent in 15 minutes. That’s a ways off, but the Silicon Valley–based company announced in March that it had begun shipping its prototype Alpha-2 semi-solid-state cells to manufacturers for testing.

Toyota is among a group of companies not looking to hedge their bets. The automaker, ignoring naysayers, aims to commercialize solid-state batteries by 2027 that it says will give an EV a range of 1,200 km on a single charge and allow 10-minute fast charging. It attributes its optimism to breakthroughs addressing durability issues. And for companies like Solid Power, it’s also solid-state or bust. Solid Power, which aims to commercialize a lithium battery with a proprietary sulfide-based solid electrolyte, has partnered with major automakers Ford and BMW. ProLogium Technology, which is also forging ahead with preparations for a solid-state battery rollout, claims that it will start delivering batteries this year that combine a ceramic oxide electrolyte with a lithium-free soft cathode (for energy density exceeding 500 Wh/kg). The company, which has teamed up with Mercedes-Benz, demonstrated confidence in its timetable by opening the world’s first giga-level solid-state lithium ceramic battery factory earlier this year in Taoyuan, Taiwan.

by Evan Ackerman on 19. June 2024. at 11:00

Insects have long been an inspiration for robots. The insect world is full of things that are tiny, fully autonomous, highly mobile, energy efficient, multimodal, self-repairing, and I could go on and on but you get the idea—insects are both an inspiration and a source of frustration to roboticists because it’s so hard to get robots to have anywhere close to insect capability.

We’re definitely making progress, though. In a paper published last month in IEEE Robotics and Automation Letters, roboticists from Shanghai Jong Tong University demonstrated the most buglike robotic bug I think I’ve ever seen.

---

A Multi-Modal Tailless Flapping-Wing Robot www.youtube.com

Okay so it may not look the most buglike, but it can do many very buggy bug things, including crawling, taking off horizontally, flying around (with six degrees of freedom control), hovering, landing, and self-righting if necessary. JT-fly weighs about 35 grams and has a wingspan of 33 centimeters, using four wings at once to fly at up to 5 meters per second and six legs to scurry at 0.3 m/s. Its 380 milliampere-hour battery powers it for an actually somewhat useful 8-ish minutes of flying and about 60 minutes of crawling.

While that amount of endurance may not sound like a lot, robots like these aren’t necessarily intended to be moving continuously. Rather, they move a little bit, find a nice safe perch, and then do some sensing or whatever until you ask them to move to a new spot. Ideally, most of that movement would be crawling, but having the option to fly makes JT-fly exponentially more useful.

Or, potentially more useful, because obviously this is still very much a research project. It does seem like there’s a bunch more optimization that could be done here. For example, JT-fly uses completely separate systems for flying and crawling, with two motors powering the legs and two additional motors powering the wings—plus two wing servos for control. There’s currently a limited amount of onboard autonomy, with an inertial measurement unit, barometer, and wireless communication, but otherwise not much in the way of useful payload.

Insects are both an inspiration and a source of frustration to roboticists because it’s so hard to get robots to have anywhere close to insect capability.

It won’t surprise you to learn that the researchers have disaster-relief applications in mind for this robot, suggesting that “after natural disasters such as earthquakes and mudslides, roads and buildings will be severely damaged, and in these scenarios, JT-fly can rely on its flight ability to quickly deploy into the mission area.” One day, robots like these will actually be deployed for disaster relief, and although that day is not today, we’re just a little bit closer than we were before.

“A Multi-Modal Tailless Flapping-Wing Robot Capable of Flying, Crawling, Self-Righting and Horizontal Takeoff,” by Chaofeng Wu, Yiming Xiao, Jiaxin Zhao, Jiawang Mou, Feng Cui, and Wu Liu from Shanghai Jong Tong University, is published in the May issue of IEEE Robotics and Automation Letters.

by Glenn Zorpette on 18. June 2024. at 19:21

Various scenarios to getting to net zero carbon emissions from power generation by 2050 hinge on the success of some hugely ambitious initiatives in renewable energy, grid enhancements, and other areas. Perhaps none of these are more audacious than an envisioned renaissance of nuclear power, driven by advanced-technology reactors that are smaller than traditional nuclear power reactors.

What many of these reactors have in common is that they would use a kind of fuel called high-assay low-enriched uranium (HALEU). Its composition varies, but for power generation, a typical mix contains slightly less than 20 percent by mass of the highly fissionable isotope uranium-235 (U-235). That’s in contrast to traditional reactor fuels, which range from 3 percent to 5 percent U-235 by mass, and natural uranium, which is just 0.7 percent U-235.

Now, though, a paper in Science magazine has identified a significant wrinkle in this nuclear option: HALEU fuel can theoretically be used to make a fission bomb—a fact that the paper’s authors use to argue for the tightening of regulations governing access to, and transportation of, the material. Among the five authors of the paper, which is titled “The Weapons Potential of High-Assay Low-Enriched Uranium,” is IEEE Life Fellow Richard L. Garwin. Garwin was the key figure behind the design of the thermonuclear bomb, which was tested in 1952.

The Science paper is not the first to argue for a reevaluation of the nuclear proliferation risks of HALEU fuel. A report published last year by the National Academies, “Merits and Viability of Different Nuclear Fuel Cycles and Technology Options and the Waste Aspects of Advanced Nuclear Reactors,” devoted most of a chapter to the risks of HALEU fuel. It reached similar technical conclusions to those of the Science article, but did not go as far in its recommendations regarding the need to tighten regulations.

Why is HALEU fuel concerning?

Conventional wisdom had it that U-235 concentrations below 20 percent were not usable for a bomb. But “we found this testimony in 1984 from the chief of the theoretical division of Los Alamos, who basically confirmed that, yes, indeed, it is usable down to 10 percent,” says R. Scott Kemp of MIT, another of the paper’s authors. “So you don’t even need centrifuges, and that’s what really is important here.”

Centrifuges arranged very painstakingly into cascades are the standard means of enriching uranium to bomb-grade material, and they require scarce and costly resources, expertise, and materials to operate. In fact, the difficulty of building and operating such cascades on an industrial scale has for decades served as an effective barrier to would-be builders of nuclear weapons. So any route to a nuclear weapon that bypassed enrichment would offer an undoubtedly easier alternative. The question now is, how much easier?

“It’s not a very good bomb, but it could explode and wreak all kinds of havoc.”

Adding urgency to that question is an anticipated gold rush in HALEU, after years of quiet U.S. government support. The U.S. Department of Energy is spending billions to expand production of the fuel, including US $150 million awarded in 2022 to a subsidiary of Centrus Energy Corp., the only private company in the United States enriching uranium to HALEU concentrations. (Outside of the United States, only Russia and China are producing HALEU in substantial quantities.) Government support also extends to the companies building the reactors that will use HALEU. Two of the largest reactor startups, Terrapower (backed in part by Bill Gates) and X-Energy, have designed reactors that run on forms of HALEU fuel, and have received billions in funding under a DOE program called Advanced Reactor Demonstration Projects.

The difficulty of building a bomb based on HALEU is a murky subject, because many of the specific techniques and practices of nuclear weapons design are classified. But basic information about the standard type of fission weapon, known as an implosion device, has long been known publicly. (The first two implosion devices were detonated in 1945, one in the Trinity test and the other over Nagasaki, Japan.) An implosion device is based on a hollow sphere of nuclear material. In a modern weapon this material is typically plutonium-239, but it can also be a mixture of uranium isotopes that includes a percentage of U-235 ranging from 100 percent all the way down to, apparently, around 10 percent. The sphere is surrounded by shaped chemical explosives that are exploded simultaneously, creating a shockwave that physically compresses the sphere, reducing the distance between its atoms and increasing the likelihood that neutrons emitted from their nuclei will encounter other nuclei and split them, releasing more neutrons. As the sphere shrinks it goes from a subcritical state, in which that chain reaction of neutrons splitting nuclei and creating other neutrons cannot sustain itself, to a critical state, in which it can. As the sphere continues to compress it achieves supercriticality, after which an injected flood of neutrons triggers the superfast, runaway chain reaction that is a fission explosion. All this happens in less than a millisecond.

The authors of the Science paper had to walk a fine line between not revealing too many details about weapons design while still clearly indicating the scope of the challenge of building a bomb based on HALEU. They acknowledge that the amount of HALEU material needed for a 15-kiloton bomb—roughly as powerful as the one that destroyed Hiroshima during the second World War—would be relatively large: in the hundreds of kilograms, but not more than 1,000 kg. For comparison, about 8 kg of Pu-239 is sufficient to build a fission bomb of modest sophistication. Any HALEU bomb would be commensurately larger, but still small enough to be deliverable “using an airplane, a delivery van, or a boat sailed into a city harbor,” the authors wrote.

They also acknowledged a key technical challenge for any would-be weapons makers seeking to use HALEU to make a bomb: preinitiation. The large amount of U-238 in the material would produce many neutrons, which would likely result in a nuclear chain reaction occurring too soon. That would sap energy from the subsequent triggered runaway chain reaction, limiting the explosive yield and producing what’s known in the nuclear bomb business as a “fizzle.“ However, “although preinitiation may have a bigger impact on some designs than others, even those that are sensitive to it could still produce devastating explosive power,” the authors conclude.

In other words, “it’s not a very good bomb, but it could explode and wreak all kinds of havoc,” says John Lee, professor emeritus of nuclear engineering at the University of Michigan. Lee was a contributor to the 2023 National Academies report that also considered risks of HALEU fuel and made policy recommendations similar to those of the Science paper.

Critics of that paper argue that the challenges of building a HALEU bomb, while not insurmountable, would stymie a nonstate group. And a national weapons program, which would likely have the resources to surmount them, would not be interested in such a bomb, because of its limitations and relative unreliability.

“That’s why the IAEA [International Atomic Energy Agency], in their wisdom, said, ‘This is not a direct-use material,’” says Steven Nesbit, a nuclear-engineering consultant and past president of the American Nuclear Society, a professional organization. “It’s just not a realistic pathway to a nuclear weapon.”

The Science authors conclude their paper by recommending that the U.S. Congress direct the DOE’s National Nuclear Security Administration (NNSA) to conduct a “fresh review” of the risks posed by HALEU fuel. In response to an email inquiry from IEEE Spectrum, an NNSA spokesman, Craig Branson, replied: “To meet net-zero emissions goals, the United States has prioritized the design, development, and deployment of advanced nuclear technologies, including advanced and small modular reactors. Many will rely on HALEU to achieve smaller designs, longer operating cycles, and increased efficiencies over current technologies. They will be essential to our efforts to decarbonize while meeting growing energy demand. As these technologies move forward, the Department of Energy and NNSA have programs to work with willing industrial partners to assess the risk and enhance the safety, security, and safeguards of their designs.”

The Science authors also called on the U.S. Nuclear Regulatory Commission (NRC) and the IAEA to change the way they categorize HALEU fuel. Under the NRC’s current categorization, even large quantities of HALEU are now considered category II, which means that security measures focus on the early detection of theft. The authors want weapons-relevant quantities of HALEU reclassified as category I, the same as for quantities of weapons-grade plutonium or highly enriched uranium sufficient to make a bomb. Category I would require much tighter security, focusing on the prevention of theft.

Nesbit scoffs at the proposal, citing the difficulties of heisting perhaps a metric tonne of nuclear material. “Blindly applying all of the baggage that goes with protecting nuclear weapons to something like this is just way overboard,” he says.

But Lee, who performed experiments with HALEU fuel in the 1980s, agrees with his colleagues. “Dick Garwin and Frank von Hipple [and the other authors of the Science paper] have raised some proper questions,” he declares. “They’re saying the NRC should take more precautions. I’m all for that.”

by Matthew S. Smith on 18. June 2024. at 16:10

Autonomous vehicles (AVs) have made headlines in recent months, though often for all the wrong reasons. Cruise, Waymo, and Tesla are all under U.S. federal investigation for a variety of accidents, some of which caused serious injury or death.

A new paper published in Nature puts numbers to the problem. Its authors analyzed over 37,000 accidents involving autonomous and human-driven vehicles to gauge risk across several accident scenarios. The paper reports AVs were generally less prone to accidents than those driven by humans, but significantly underperformed humans in some situations.

“The conclusion may not be surprising given the technological context,” said Shengxuan Ding, an author on the paper. “However, challenges remain under specific conditions, necessitating advanced algorithms and sensors and updates to infrastructure to effectively support AV technology.”

The paper, authored by two researchers at the University of Central Florida, analyzed data from 2,100 accidents involving advanced driving systems (SAE Level 4) and advanced driver-assistance systems (SAE Level 2) alongside 35,113 accidents involving human-driven vehicles. The study pulled from publicly available data on human-driven vehicle accidents in the state of California and the AVOID autonomous vehicle operation incident dataset, which the authors made public last year.

While the breadth of the paper’s data is significant, the paper’s “matched case-control analysis” is what sets it apart. Autonomous and human-driven vehicles tend to encounter different roads in different conditions, which can skew accident data. The paper categorizes risks by the variables surrounding the accident, such as whether the vehicle was moving straight or turning, and the conditions of the road and weather.

Level 4 self-driving vehicles were roughly 36 percent less likely to be involved in moderate injury accidents and 90 percent less likely to be involved in a fatal accident.

SAE Level 4 self-driving vehicles (those capable of full self-driving without a human at the wheel) performed especially well by several metrics. They were roughly 36 percent less likely to be involved in moderate injury accidents and 90 percent less likely to be involved in a fatal accident. Compared to human-driven vehicles, the risk of rear-end collision was roughly halved, and the risk of a broadside collision was roughly one-fifth. Level 4 AVs were close to one-fifthtieth as likely to run off the road.

The paper’s findings are generally favorable for level 4 AVs, but they perform worse in turns, and at dawn and dusk.Nature

These figures look good for AVs. However, Missy Cummings, director of George Mason University’s Autonomy and Robotics Center and former safety advisor for the National Highway Traffic Safety Administration, was skeptical of the findings.

“The ground rules should be that when you analyze AV accidents, you cannot combine accidents with self-driving cars [SAE Level 4] with the accidents of Teslas [SAE Level 2],” said Cummings. She took issue with discussing them in tandem and points out these categories of vehicles operate differently—so much so that Level 4 AVs aren’t legal in every state, while Level 2 AVs are.

Mohamed Abdel-Aty, an author on the paper and director of the Smart & Safe Transportation Lab at the University of Central Florida, said that while the paper touches on both levels of autonomy, the focus was on Level 4 autonomy. “The model which is the main contribution to this research compared only level 4 to human-driven vehicles,” he said.

And while many findings were generally positive, the authors highlighted two significant negative outcomes for level 4 AVs. It found they were over five times more likely to be involved in an accident at dawn and dusk. They were relatively bad at navigating turns as well, with the odds of an accident during a turn almost doubled compared to those for human-driven vehicles.

More data required for AVs to be “reassuring”

The study’s finding of higher accident rates during turns and in unusual lighting conditions highlight two major categories of challenges facing self-driving vehicles: intelligence and data.

J. Christian Gerdes, codirector of the Center for Automotive Research at Stanford University, said turning through traffic is among the most demanding situations for an AV’s artificial intelligence. “That decision is based a lot on the actions of other road users around you, and you’re going to make the choice based on what you predict.”

Cummings agreed with Gerdes. “Any time uncertainty increases [for an AV], you’re going to see an increased risk of accident. Just by the fact you’re turning, that increases uncertainty, and increases risk.”

AVs’ dramatically higher risk of accidents at dawn and dusk, on the other hand, points towards issues with the data captured by a vehicle’s sensors. Most AVs use a combination of radar and visual sensor systems, and the latter is prone to error in difficult lighting.

It’s not all bad news for sensors, though. Level 4 AVs were drastically better in rain and fog, which suggests that the presence of radar and lidar systems gives AVs an advantage in weather conditions that reduce visibility. Gerdes also said AVs, unlike humans, don’t tire or become distracted when driving through weather that requires more vigilance.

While the paper found AVs have a lower risk of accident overall, that doesn’t mean they’ve passed the checkered flag. Gerdes said poor performance in specific scenarios is meaningful and should rightfully make human passengers uncomfortable.

“It’s hard to make the argument that [AVs] are so much safer driving straight, but if [they] get into other situations, they don’t do as well. People will not find that reassuring,” said Gerdes.

The relative lack of data for Level 4 systems is another barrier. Level 4 AVs make up a tiny fraction of all vehicles on the road and only operate in specific areas. AVs are also packed with sensors and driven by an AI system that may make decisions for a variety of reasons that remain opaque in accident data.

While the paper accounts for the low total number of accidents in its statistical analysis, the authors acknowledge more data is necessary to determine the precise cause of accidents, and hope their findings will encourage others to assist. “I believe one of the benefits of this study is to draw the attention of authorities to the need for better data,” said Ding.

On that, Cummings agreed. “We do not have enough information to make sweeping statements,” she said.

by G. Pascal Zachary on 18. June 2024. at 10:00

In the summer of 1945, Robert J. Oppenheimer and other key members of the Manhattan Project gathered in New Mexico to witness the first atomic bomb test. Among the observers was Vannevar Bush, who had overseen the Manhattan Project and served as the sole liaison to U.S. President Franklin D. Roosevelt on progress toward the bomb.

Remarkably, given his intense wartime responsibilities, Bush continued to develop his own ideas about computing and information. Just days before the Trinity test, he had published in The Atlantic Monthly a futuristic account of networks of information knitted together via “associative trails”—which we would now call hypertext or hyperlinks. To this day, Bush’s article—titled “As We May Think”—and his subsequent elaborations of networked information appliances are credited with shaping what would become the personal computer and the World Wide Web. And during his lifetime, Bush was celebrated as one of the nation’s leading prophets of technological change and the most influential proponent of government funding of science and engineering.

Vannevar Bush’s influential 1945 essay “As We May Think” shaped the subsequent development of the personal computer and the World Wide Web. The Atlantic Monthly

And yet, if you watched this year’s Oscar-winning Oppenheimer, Bush is only a minor character. Played by actor Matthew Modine, he testifies before a secret government panel that will decide whether Oppenheimer, scientific director of the Manhattan Project, should be stripped of his security clearance and banished from participating in future government decisions on sensitive technological issues.

“Try me, if you want to try him,” Bush defiantly tells the panel. Alas, tragedy unfolds when the panel punishes Oppenheimer for his opposition to testing the nation’s first hydrogen bomb. No more is said about Bush, even though he also opposed the first H-bomb test, on the grounds that the test, held on 1 November 1952, would help the Soviet Union build its own superweapon and accelerate a nuclear arms race. Bush was spared sanction and continued to serve in government, while Oppenheimer became a pariah.

Today, though, Oppenheimer is lionized while Bush is little known outside a small circle of historians, computer scientists, and policy thinkers. And yet, Bush’s legacy is without a doubt the more significant one for engineers and scientists, entrepreneurs, and public policymakers. He died at the age of 84 on 28 June 1974, and the 50th anniversary of his death seems like a good time to reflect on all that Vannevar Bush did to harness technological innovation as the chief source of economic, political, and military power for the United States and other leading nations.

Vannevar Bush and the Funding of Science & Engineering

Beginning in 1940, and with the ear of the president and leading scientific and engineering organizations, Vannevar Bush promoted the importance of supporting all aspects of research, including in universities, the military, and industry. Bush’s vision was shaped by World War II and America’s need to rapidly mobilize scientists and engineers for war fighting and defense. And it deepened during the long Cold War.

Bush’s pivotal contribution was his creation of the “research contract,” whereby public funds are awarded to civilian scientists and engineers based on effort, not just outcomes (as had been normal before World War II). This freedom to try new things and take risks transformed relations between government, business, and academia. By the end of the war, Bush’s research organization was spending US $3 million a week (about $52 million in today’s dollars) on some 6,000 researchers, most of them university professors and corporate engineers.

On its 3 April 1944 cover, Time called Vannevar Bush the “General of Physics,” for his role in accelerating wartime R&D.Ernest Hamlin Baker/TIME

Celebrated as the “general of physics” on the cover of Time magazine in 1944, Bush served as the first research chief of the newly created Department of Defense in 1947. Three years later, he successfully advocated for the creation of a national science foundation, to nourish and sustain civilian R&D. In launching his campaign for the foundation, Bush issued a report, entitled Science, The Endless Frontier, in which he argued that the nation’s future prosperity and the American spirit of “frontier” exploration depended on advances in science and engineering.

Bush’s influence went well beyond the politics of research and the mobilization of technology for national security. He was also a business innovator. In the 1920s, he cofounded Raytheon, and the company competed with behemoth RCA in the design and manufacture of vacuum tubes. As a professor and later dean of engineering at the Massachusetts Institute of Technology, he crafted incentives for professors to consult part time for business, setting in motion in the 1920s and 1930s practices now considered essential to science-based industry.

Bush’s beliefs influenced Frederick Terman, a doctoral student of his, to join Stanford University, where Terman played a decisive role in the birth of Silicon Valley. Another Bush doctoral student, Claude Shannon, joined Bell Labs and founded information theory. As a friend and trusted adviser to Georges Doriot, Bush helped launch one of the first venture capital firms, American Research and Development Corp.

Vannevar Bush’s Contributions to Computing

Starting in the 1920s, Bush began designing analog computing machines, known as differential analyzers. This version was at Aberdeen Proving Ground, in Maryland.MIT Museum

But wait, there’s more! Bush was a major figure in the early history of modern computing. In the 1930s, he gained prestige as the designer of a room-size analog computing machine known as the “differential analyzer,” then considered the most powerful calculating machine on the planet. It was visually impressive enough that UCLA’s differential analyzer had a major cameo in the 1951 sci-fi movie When Worlds Collide.

In the 1940s, despite his busy schedule with the Manhattan Project, Bush set aside time to envision and build working models of a desktop “memory extender,” or memex, to assist professionals in managing information and making decisions. And, as mentioned, he published that pivotal Atlantic article.

For engineers, Bush carries a special significance because of his passionate arguments throughout his life that all engineers—especially electrical engineers—deserve the same professional status as doctors, lawyers, and judges. Before World War II, engineers were viewed chiefly as workers for hire who did what they were told by their employers, but Bush eloquently insisted that engineers possessed professional rights and obligations and that they delivered their expert judgments independently and, when feasible, with the public interest in mind.

Vannevar Bush considered engineering not just a job but a calling. John Lent/AP

From the distance of a half century, Bush’s record as a futurist was mixed. He failed to envision the enormous expansion of both digital processing power and storage. He loudly proclaimed that miniaturized analog images stored on microfilm would long provide ample storage. (To be fair, many old microfilm and microfiche archives remain readable, unlike, say, digital video disks and old floppies.)

And yet, Bush’s ideas about the future of information have proved prescient. He believed, for example, that human consciousness could be enhanced through computational aids and that the automation of routine cognitive tasks could liberate human minds to concentrate and solve more difficult problems.

In this regard, Bush prefigures later computing pioneers like Douglas Engelbart (inventor of the mouse) and Larry Page (cofounder of Google), who promoted the concept of human “augmentation” through innovative digital means, such as hypertext and search, and enhancing the speed, accuracy, and depth of purposeful thought. Indeed, today’s debate over the harm to humans from generative AI could benefit from Bush’s own calm assessment about the creative, intellectual, and artistic benefits to be gained from “the revolution in machines to reduce mental drudgery.” The subject of human enhancement through digital systems was “almost constantly” on his mind, he wrote in his 1970 memoir, Pieces of the Action, four years before his death. Bush cautioned against hysteria in the face of digitally mediated cognitive enhancements. And he insisted that our technological systems should maintain the proverbial “human in the loop,” in order to honor and safeguard our values in the tricky management of digital information systems.

The fate of human culture and values was not Bush’s only worry. In his later life, he fretted about the spread of nuclear weapons and the risk of their use. Fittingly, as the titular head of the Manhattan Project and, in the 1950s, an opponent of testing the first H-bomb, he saw nuclear weapons as an existential threat to all life on the planet.

Bush identified no ultimate solutions to these problems. Having done so much to enhance and solidify the role of scientists and engineers in the advancement of society, he nevertheless foresaw an uncertain world, where scientific and technological outcomes would also continue to challenge us.

by Emily Waltz on 17. June 2024. at 13:00

Inside a shipping container in an industrial area of Venice, the Italian startup 9-Tech is taking a crack at a looming global problem: how to responsibly recycle the 54 million to 160 million tonnes of solar modules that are expected to reach the end of their productive lives by 2050. Recovering the materials won’t be easy. Solar panels are built to withstand any environment on Earth for 20 to 30 years, and even after sitting in the sun for three decades, the hardware is difficult to dismantle. In fact, most recycling facilities trash the silicon, silver, and copper—the most valuable but least accessible materials in old solar panels—and recover only the aluminum frames and glass panes.

The startup 9-Tech operates its pilot plant out of a modified shipping container housed at the Green Propulsion Laboratory in the industrial port of Marghera in Venice.Luigi Avantaggiato

The need for recycling will only grow as the world increasingly deploys solar power. More than 1.2 terawatts of solar power has already been deployed globally. Solar panels are currently being distributed at a rate of more than 400 gigawatts per year, and the rate is expected to increase to a whopping 3 TW per year by 2030, according to a literature analysis by researchers at the National Renewable Energy Laboratory (NREL).

In an attempt to stop a mountain of photovoltaic garbage from accumulating, researchers are pursuing better recycling methods. The most advanced methods proposed so far can recover at least 90 percent of the copper, silver, silicon, glass, and aluminum in a crystalline silicon PV module. But these processes are expensive and often involve toxic chemicals. No recycling method has proven to be as cheap as landfilling, and very few operate on an industrial scale, says Garvin Heath, principal environmental engineer at NREL, who manages a group of international experts assigned by the International Energy Agency to analyze PV sustainability.

The founders of 9-Tech say they have a better way. Their process is a noisy one involving a combustion furnace, an ultrasound bath, and mechanical sorting, the vibrations of which shake the floor of the modest freight container where they have been testing their operation for nearly two years. The company uses no toxic chemicals, releases no pollutants into the environment, and recovers up to 90 percent of the materials in a solar panel, says Francesco Miserocchi, chief technology officer at 9-Tech.

Bits of silicon and glass are separated from the rest of the panel. Luigi Avantaggiato

How to Recycle Solar Panels

After the frame, glass, and junction box are removed from a PV panel, the inner, bendable layers of silicon, polymers, and metal conductors remain. Workers cut the inner layers into large sections in preparation for the oven.Luigi Avantaggiato

The company tailors its process to crystalline silicon solar panels, which make up 97 percent of the global PV market. The panels typically consist of an array of silicon wafers doped with boron and phosphorus, and topped with an antireflective coating of silicon nitride. Silver conductors are screen printed onto the wafer surface, and copper conductors are soldered onto the array in a grid pattern. To protect the materials from moisture and damage, manufacturers laminate the entire array in adhesive polymers—usually ethylene-vinyl acetate. Then they encase the laminated array in sheets of tempered glass, frame the whole thing in aluminum, seal the edges, and attach a junction box on the back.

When it’s time to recycle a panel, one of the most challenging steps is removing the polymers, which stick to everything. “It’s not just the edges or a couple of dots of glue. It’s an entire surface—several square feet—of polymer,” says Heath. The polymer can be burned off, but this releases carbon monoxide, hydrofluoric acid, and other harmful pollutants. Separating the silver conductors also proves challenging because they’re applied in a very thin layer–about 10 to 20 micrometers–that is strongly attached to the silicon. Removing them typically involves toxic reagents such as hydrofluoric acid, nitric acid, or sodium hydroxide.

The team at 9-Tech addresses these challenges in two ways. They recover the silver using ultrasound rather than toxic chemicals, and although they burn the polymers, they capture the pollutants emitted.

Layers of silicon and polymer are fed into a continuous combustion furnace, which heats the materials to over 400 °C, vaporizing the polymers. 9-Tech

The process at the company’s pilot plant starts with workers manually removing the aluminum frame, junction box, and tempered glass. This leaves a sandwich of polymers, silicon wafers, and metal conductors. Without the frame or glass, the sandwich layers bend, shattering the fragile silicon into small pieces. Workers crack the tempered glass and then send all the materials, which are mostly still in place because of the polymers, into a continuous combustion furnace. Heated to over 400 °C, the polymers vaporize, and a filter captures the pollutants. The system also captures the heat from the furnace and reuses it for energy efficiency.

A mechanical roller separates the copper grid after the PV materials exit the furnace.Luigi Avantaggiato

After the materials exit the oven, mechanical sieves separate the copper, glass, and silicon. Top: Luigi Avantaggiato; Bottom: 9-Tech

As the remaining material exits the furnace, a roller mechanically strips out the copper. A series of sieves sort the broken bits of glass and silicon based on thickness. The silicon pieces, still laced with silver, are immersed in a bath of organic acid and treated with ultrasound to loosen the bonds between the elements. The ultrasound works by propagating sound waves into the acid bath, resulting in alternating high- and low-pressure cycles. If the waves are intense enough, they create cavitation bubbles that mechanically interact with the material, causing the silver to dislodge from the silicon, explains Pietrogiovanni Cerchier, CEO at 9-Tech.

Finally, workers remove the silicon fragments from the ultrasound bath with a mesh net. This leaves a fine silver dust in the solution, which can be recovered by filtration or centrifuge. All told, Cerchier says, 9-Tech’s pilot plant can recover 90 percent of the silver, 95 percent of the silicon, and 99 percent or more of the copper, aluminum, and glass from a PV module. What’s more, the material is considered highly pure, which increases the types of applications for which it can be reused.

Workers at 9-Tech immerse silver-laced silicon pieces in a bath of organic acid and treat it with ultrasound to loosen the bonds between the elements. 9-Tech

The startup’s recycling process is more expensive than existing methods that recover only aluminum and glass. But the extraction of high-purity silicon, silver, and copper should offset the extra cost, Miserocchi says. Plus, it’s more efficient than mining for virgin elements. You can extract about 500 grams of silver from a tonne of solar panels, but only 165 grams of silver from a tonne of ore, he says. “A photovoltaic panel at the end of its life still has a lot to give,” says Miserocchi. “It can be considered a small mine of precious elements.”

Silver emerges from the silicon bath in a fine dust. 9-Tech

Dozens of New Ways to Recycle PV Panels

High-purity copper, glass, and silicon are recovered from 9-Tech’s PV-panel recycling process.Luigi Avantaggiato

The 9-Tech team will know more about the profitability of their method after they build a larger demonstration facility over the next 18 months. That plant, to be located in the same industrial district of Venice as the shipping container, will be able to handle up to 800 solar modules a day. Their pilot plant processes only about seven modules a day.

The company’s approach is one of many recycling methods for crystalline silicon PV panels in development. A comprehensive review published in April in the Journal of Cleaner Production identified dozens of other efforts globally, including thermal, chemical, mechanical, and optical approaches. The most common method involves grinding the silicon, metal, and polymer layers into small pieces, separating them by density, and recovering the silicon and metal with a thermal or chemical process. Other processes include laser irradiation, high-voltage pulses, optical sorting, pyrolysis, chemical solvents, etching, and delaminating with a hot knife.

Driving this innovation, in part, are regulations adopted by the European Union in 2012. The rules require all PV panel manufacturers in the EU to run take-back or recycling programs, or partner with other recycling schemes. As a result, Germany, which has the most solar power capacity in Europe, has one of the largest PV recycling systems in the world.

“A photovoltaic panel at the end of its life still has a lot to give,” says Miserocchi. “It can be considered a small mine of precious elements.”

But recycling is a high-volume business, and aside from catastrophic weather events that wipe out solar power stations, spent solar modules reach recyclers at a relative trickle. And then there’s the challenge of finding a second life for the materials after they are recovered—a supply chain that’s not well developed.

First Solar, a global PV manufacturer based in Tempe, Ariz., addressed both of these issues on a large scale by building an in-house recycling program with seven facilities in five countries. The company makes cadmium telluride thin-film solar panels that buyers can purchase with the recycling price built in. At the end of the panels’ lives, buyers send them back to First Solar for recycling into new products. The semiconductor material can be recycled up to 41 times, giving it a life-span of more than 1,200 years, according to the company. But the glass isn’t pure enough to be reused in solar modules, so the company plans to supply it to float-glass manufacturers for use in windows and doors.

The challenges with recycling have inspired researchers to rethink the way crystalline solar panels are made. For example, some manufacturers are trying to reduce or eliminate the difficult-to-recover silver, replacing it with other conductive metals. And a team at NREL demonstrated in February a way to eliminate polymers in PV panels by laser welding the glass panes instead, which may do a better job sealing out moisture. That technique may lend itself to perovskite solar modules, a promising technology that is particularly susceptible to moisture and corrosion.

“Recycling shouldn’t be the only strategy,” says Heath. People should consider alternative ways to repair or reuse solar panels to extend their lives before resorting to recycling, he says.

Additional reporting by Luigi Avantaggiato

by Ernie Smith on 16. June 2024. at 14:00

A version of this post originally appeared on Tedium, Ernie Smith’s newsletter, which hunts for the end of the long tail.

Last month, Google announced some big changes to its search engine that are, in a word, infuriating, to users like myself.

Google has started adding AI overviews to many of its search results, which essentially generate pre-processed answers to search queries. If you’re using Google to actually find websites rather than get answers, it $!@(&!@ sucks.

But in the midst of all this, Google quietly added something else to its results—a “Web” filter that presents what Google used to look like a decade ago, no extra junk. While Google made its AI-focused changes known on its biggest stage—during its Google I/O event—the Web filter was curiously announced on Twitter by Search Liaison Danny Sullivan.

As Sullivan wrote:

We’ve added this after hearing from some that there are times when they’d prefer to just see links to web pages in their search results, such as if they’re looking for longer-form text documents, using a device with limited internet access, or those who just prefer text-based results shown separately from search features. If you’re in that group, enjoy!

The results are fascinating. It’s essentially Google, minus the extra fluff. No parsing of the information in the results. No surfacing metadata like address or link info. No knowledge panels, but also, no ads. It looks like the Google we learned to love in the early 2000s, buried under the “More” menu.

This is what Google search used to look like, without any extras, and it can look like that again.Ernie Smith

For power searchers like myself, it’s likely going to be an amazing tool. But Google’s decision to bury it ensures that few people will use it. The company has essentially bet that you’ll be better off with a pre-parsed guess produced by its AI engine.

It’s worth understanding the tradeoffs, though. A simplified view does not replace the declining quality of Google’s results, largely caused by decades of SEO optimization by website creators. The same overly optimized results are going to be there, like it or not. It is not Google circa 2001; it is a Google-circa-2001 presentation of Google circa 2024, a very different site.

But if you understand the tradeoffs, it can be a great tool.

And here’s the trick to using it without having to click the ‘Web’ option buried in a menu every single time. Google does not make it easy, but by adding a URL parameter to your search—in this case, “udm=14”—you can get directly to the Web results in a search.

That sounds like extra work until you realize that many browsers allow you to add custom search engines by adding the %s entry as a stand-in for the search term you put in. And it works great in the case of Google.

You can specify the default url for the omnibar search on your browser. Ernie Smith

In Vivaldi, my weapon of choice, I did this:

●Go to Settings -> Search

●Look at the list of search engines, and hit the plus button at the bottom left of the dialog box to add a new one

●Name the new item “Google Web Only,” and give it the nickname of “gw”

●Set the URL as https://www.google.com/search?q=%s&udm=14

●Set it as your default search

Now, when you use the omnibar on your browser of choice, it will automatically push you to the Google Web Only search. If you want a more traditional search, add a “g” in front of the search in your omnibar, and it will give you the full-fat search, knowledge panels and all. Don’t want to make it your default? Don’t.

A variant of this should work for most Chromium-based browsers, including Chrome proper. It is also possible in Firefox with an extension. Safari, which does not allow you to add custom search engines by default, is a little more complicated, but it is possible through the use of custom extensions like HyperWeb for iOS. I’m still looking for a Safari-for-Mac solution.

Or, you can use a front-end that I created at udm14.com or udm14.org.

When you want something more elemental, less adulterated, it’s there, no extra junk.

It’s depressing that it’s gotten to this, isn’t it?

by Evan Ackerman on 14. June 2024. at 17:12

Video Friday is your weekly selection of awesome robotics videos, collected by your friends at IEEE Spectrum robotics. We also post a weekly calendar of upcoming robotics events for the next few months. Please send us your events for inclusion.

RoboCup 2024: 17–22 July 2024, EINDHOVEN, NETHERLANDS

ICRA@40: 23–26 September 2024, ROTTERDAM, NETHERLANDS

IROS 2024: 14–18 October 2024, ABU DHABI, UAE

ICSR 2024: 23–26 October 2024, ODENSE, DENMARK

Cybathlon 2024: 25–27 October 2024, ZURICH

Enjoy today’s videos!

There’s a Canadian legend about a flying canoe, because of course there is. The legend involves drunkenness, a party with some ladies, swearing, and a pact with the devil, because of course it does. Fortunately for the drone in this video, it needs none of that to successfully land on this (nearly) flying canoe, just some high-friction shock absorbing legs and judicious application of reverse thrust.

[ Createk ]

Thanks, Alexis!

This paper summarizes an autonomous driving project by musculoskeletal humanoids. The musculoskeletal humanoid, which mimics the human body in detail, has redundant sensors and a flexible body structure. We reconsider the developed hardware and software of the musculoskeletal humanoid Musashi in the context of autonomous driving. The respective components of autonomous driving are conducted using the benefits of the hardware and software. Finally, Musashi succeeded in the pedal and steering wheel operations with recognition.

[ Paper ] via [ JSK Lab ]

Thanks, Kento!

Robust AI has been kinda quiet for the last little while, but their Carter robot continues to improve.

[ Robust AI ]

One of the key arguments for building robots that have similar form factors to human beings is that we can leverage the massive human data for training. In this paper, we introduce a full-stack system for humanoids to learn motion and autonomous skills from human data. We demonstrate the system on our customized 33-degrees-of-freedom 180 centimeter humanoid, autonomously completing tasks such as wearing a shoe to stand up and walk, unloading objects from warehouse racks, folding a sweatshirt, rearranging objects, typing, and greeting another robot with 60-100 percent success rates using up to 40 demonstrations.

[ HumanPlus ]

We present OmniH2O (Omni Human-to-Humanoid), a learning-based system for whole-body humanoid teleoperation and autonomy. Using kinematic pose as a universal control interface, OmniH2O enables various ways for a human to control a full-sized humanoid with dexterous hands, including using real-time teleoperation through VR headset, verbal instruction, and RGB camera. OmniH2O also enables full autonomy by learning from teleoperated demonstrations or integrating with frontier models such as GPT-4.

[ OmniH2O ]

A collaboration between Boxbot, Agility Robotics, and Robust.AI at Playground Global. Make sure and watch until the end to hear the roboticists in the background react when the demo works in a very roboticist way.

::clap clap clap:: yaaaaayyyyy....

[ Robust AI ]

The use of drones and robotic devices threatens civilian and military actors in conflict areas. We started trials with robots to see how we can adapt our HEAT (Hostile Environment Awareness Training) courses to this new reality.

[ CSD ]

Thanks, Ebe!

How to make humanoids do versatile parkour jumping, clapping dance, cliff traversal, and box pick-and-move with a unified RL framework? We introduce WoCoCo: Whole-body humanoid Control with sequential Contacts

[ WoCoCo ]

A selection of excellent demos from the Learning Systems and Robotics Lab at TUM and the University of Toronto.

[ Learning Systems and Robotics Lab ]

Harvest Automation, one of the OG autonomous mobile robot companies, hasn’t updated their website since like 2016, but some videos just showed up on YouTube this week.

[ Harvest Automation ]

Northrop Grumman has been pioneering capabilities in the undersea domain for more than 50 years. Now, we are creating a new class of uncrewed underwater vehicles (UUV) with Manta Ray. Taking its name from the massive “winged” fish, Manta Ray will operate long-duration, long-range missions in ocean environments where humans can’t go.

[ Northrop Grumman ]

Akara Robotics’ autonomous robotic UV disinfection demo.

[ Akara Robotics ]

Scientists have computationally predicted hundreds of thousands of novel materials that could be promising for new technologies—but testing to see whether any of those materials can be made in reality is a slow process. Enter A-Lab, which uses robots guided by artificial intelligence to speed up the process.

[ A-Lab ]

We wrote about this research from CMU a while back, but here’s a quite nice video.

[ CMU RI ]

Aw yiss pick and place robots.

[ Fanuc ]

Axel Moore describes his lab’s work in orthopedic biomechanics to relieve joint pain with robotic assistance.

[ CMU ]

The field of humanoid robots has grown in recent years with several companies and research laboratories developing new humanoid systems. However, the number of running robots did not noticeably rise. Despite the need for fast locomotion to quickly serve given tasks, which require traversing complex terrain by running and jumping over obstacles. To provide an overview of the design of humanoid robots with bioinspired mechanisms, this paper introduces the fundamental functions of the human running gait.

[ Paper ]

by Julianne Pepitone on 13. June 2024. at 18:00

When Yifeng Chen was a teenager in Shantou, China, in the early 2000s, he saw a TV program that amazed him. The show highlighted rooftop solar panels in Germany, explaining that the panels generated electricity to power the buildings and even earned the owners money by letting them sell extra energy back to the electricity company.

Yifeng Chen

Employer

Trina Solar

Title

Assistant vice president of technology

Member Grade

Member

Alma Maters

Sun Yat-sen University, in Guangzhou, China, and Leibniz University Hannover, in Germany

An incredulous Chen marveled at not only the technology but also the economics. A power authority would pay its customers?

It sounded like magic: useful and valuable electricity extracted from simple sunlight. The wonder of it all has fueled his dreams ever since.

In 2013 Chen earned a Ph.D. in photovoltaic sciences and technologies, and today he’s assistant vice president of technology at China’s Trina Solar, a Changzhou-based company that is one of the largest PV manufacturers in the world. He leads the company’s R&D group, whose efforts have set more than two dozen world records for solar power efficiency and output.

For Chen’s contributions to the science and technology of photovoltaic energy conversion, the IEEE member received the 2023 IEEE Stuart R. Wenham Young Professional Award from the IEEE Electron Devices Society.

“I was quite surprised and so grateful” to receive the Wenham Award, Chen says. “It’s a very high-level recognition, and there are so many deserving experts from around the world.”

Trina Solar’s push for more efficient hardware

Today’s commercial solar panels typically achieve about 20 percent efficiency: They can turn one-fifth of captured sunlight into electricity. Chen’s group is trying to make the panels more efficient.

The group is focusing on optimizing solar cell designs, including the passivated emitter and rear cell (PERC), which is the industry standard for commodity solar panels.

Invented in 1983, PERCs are used today in nearly 90 percent of solar panels on the market. They incorporate coatings on the front and back to capture sunlight more effectively and to avoid losing energy, both at the surfaces and as the sunlight travels through the cell. The coatings, known as passivation layers, are made from materials such as silicon nitride, silicon dioxide, and aluminum oxide. The layers keep negatively charged free electrons and positively charged electron holes apart, preventing them from combining at the surface of the solar cell and wasting energy.

Chen and his team have developed several ways to boost the performance of PERC panels, hitting a record of 24.5 percent efficiency in 2022. One of the technologies is a multilayer antireflective coating that helps solar panels trap more light. They also created extremely fine metallization fingers—narrow lines on solar cells’ surfaces—to collect and transport the electric current and help capture more sunlight. And they developed an advanced method for laying the strips of conductive metal that run across the solar cell, known as bus bars.

Experts predict the maximum efficiency of PERC technology will be reached soon, topping out at about 25 percent.

IEEE Member Yifeng Chen displays an i-TOPCon solar module, which has a production efficiency of more than 23 percent and a power output of up to 720 watts.Trina Solar

“So the question is: How do we get solar cells even more efficient?” Chen says.

During the past few years he and his group have been working on tunnel oxide passivated contact (TOPCon) technology. A TOPCon cell uses a thin layer of “tunneling oxide” insulating material—typically silicon dioxide—which is applied to the solar cell’s surface. Similar to the passivation layers on PERC cells, the tunnel oxide stops free electrons and electron holes from combining and wasting energy.

In 2022 Trina created a TOPCon-type panel with a record 25.5 percent efficiency, and two months ago the company announced it had achieved a record 740.6 watts for a mass-produced TOPCon solar module. The latter was the 26th record Trina set for solar power–related efficiencies and outputs.

To achieve that record-breaking performance for their TOPCon panels, Chen and his team optimized the company’s manufacturing processes including laser-induced firing, in which a laser heats part of the solar cell and creates bonds between the metal contacts and the silicon wafer. The resulting connections are stronger and better aligned, enhancing efficiency.

“We’re trying to keep improving things to trap just a little bit more sunlight,” Chen says. “Gaining 1 or 2 percent more efficiency is huge. These may sound like very tiny increases, but at scale these small improvements create a lot of value in terms of economics, sustainability, and value to society.”

As the efficiency of solar cells rises and prices drop, Chen says, he expects solar power to continue to grow around the world. China currently leads the world in installed solar power capacity, accounting for about 40 percent of global capacity. The United States is a distant second, with 12 percent, according to a 2023 Rystad Energy report. The report predicts that China’s 500 gigawatts of solar capacity in 2023 is likely to exceed 1 terawatt by 2026.

“I’m inspired by using science to create something useful for human beings, and then driven by the pursuit for excellence,” Chen says. “We can always learn something new to make that change, improve that piece of technology, and get just that little bit better.”

Trained by solar-power pioneers

Chen attended Sun Yat-sen University in Guangzhou, China, earning a bachelor’s degree in optics sciences and technologies in 2008. He stayed on to pursue a Ph.D. in photovoltaics sciences and technologies. His research was on high-efficiency solar cells made from wafer-based crystalline silicon. His adviser was Hui Shen, a leading PV professor and founder of the university’s Institute for Solar Energy Systems. Chen calls him “the first of three very important figures in my scientific career.”

In 2011 Chen spent a year as a Ph.D. student at Leibniz University Hannover, in Germany. There he studied under Pietro P. Altermatt, the second influential figure in his career.

Altermatt—a prominent silicon solar-cell expert who would later become principal scientist at Trina—advised Chen on his computational techniques for modeling and analyzing the behavior of 2D and 3D solar cells. The models play a key role in designing solar cells to optimize their output.

“Gaining 1 or 2 percent more efficiency is huge. These may sound like very tiny increases, but at scale, these small improvements create a lot of value in terms of economics, sustainability, and value to society.”

“Dr. Altermatt changed how I look at things,” Chen says. “In Germany, they really focus on device physics.”

After completing his Ph.D., Chen became a technical assistant at Trina, where he met the third highly influential person in his career: Pierre Verlinden, a pioneering photovoltaic researcher who was the company’s chief scientist.

At Trina, Chen quickly ascended through R&D roles. He has been the company’s assistant vice president of technology since 2023.

IEEE’s “treasure” trove of research

Chen joined IEEE as a student because he wanted to attend the IEEE Photovoltaic Specialists Conference, the longest-running event dedicated to photovoltaics, solar cells, and solar power.

The membership was particularly beneficial during his Ph.D. studies, he says, because he used the IEEE Xplore Digital Library to access archival papers.

“My work has certainly been inspired by papers I found via IEEE,” Chen says. “Plus, you end up clicking around and reading other work that isn’t related to your field but is so interesting.

“The publication repository is a treasure. It’s eye-opening to see what’s going on inside and outside of your industry, with new discoveries happening all the time.”

by Eliza Strickland on 13. June 2024. at 13:00

Last month when Google introduced its new AI search tool, called AI Overviews, the company seemed confident that it had tested the tool sufficiently, noting in the announcement that “people have already used AI Overviews billions of times through our experiment in Search Labs.” The tool doesn’t just return links to Web pages, as in a typical Google search, but returns an answer that it has generated based on various sources, which it links to below the answer. But immediately after the launch users began posting examples of extremely wrong answers, including a pizza recipe that included glue and the interesting fact that a dog has played in the NBA.

Renée DiResta has been tracking online misinformation for many years as the technical research manager at Stanford’s Internet Observatory.

While the pizza recipe is unlikely to convince anyone to squeeze on the Elmer’s, not all of AI Overview’s extremely wrong answers are so obvious—and some have the potential to be quite harmful. Renée DiResta has been tracking online misinformation for many years as the technical research manager at Stanford’s Internet Observatory and has a new book out about the online propagandists who “turn lies into reality.” She has studied the spread of medical misinformation via social media, so IEEE Spectrum spoke to her about whether AI search is likely to bring an onslaught of erroneous medical advice to unwary users.

I know you’ve been tracking disinformation on the Web for many years. Do you expect the introduction of AI-augmented search tools like Google’s AI Overviews to make the situation worse or better?

Renée DiResta: It’s a really interesting question. There are a couple of policies that Google has had in place for a long time that appear to be in tension with what’s coming out of AI-generated search. That’s made me feel like part of this is Google trying to keep up with where the market has gone. There’s been an incredible acceleration in the release of generative AI tools, and we are seeing Big Tech incumbents trying to make sure that they stay competitive. I think that’s one of the things that’s happening here.

We have long known that hallucinations are a thing that happens with large language models. That’s not new. It’s the deployment of them in a search capacity that I think has been rushed and ill-considered because people expect search engines to give them authoritative information. That’s the expectation you have on search, whereas you might not have that expectation on social media.

There are plenty of examples of comically poor results from AI search, things like how many rocks we should eat per day [a response that was drawn for an Onion article]. But I’m wondering if we should be worried about more serious medical misinformation. I came across one blog post about Google’s AI Overviews responses about stem-cell treatments. The problem there seemed to be that the AI search tool was sourcing its answers from disreputable clinics that were offering unproven treatments. Have you seen other examples of that kind of thing?

DiResta: I have. It’s returning information synthesized from the data that it’s trained on. The problem is that it does not seem to be adhering to the same standards that have long gone into how Google thinks about returning search results for health information. So what I mean by that is Google has, for upwards of 10 years at this point, had a search policy called Your Money or Your Life. Are you familiar with that?

I don’t think so.

DiResta: Your Money or Your Life acknowledges that for queries related to finance and health, Google has a responsibility to hold search results to a very high standard of care, and it’s paramount to get the information correct. People are coming to Google with sensitive questions and they’re looking for information to make materially impactful decisions about their lives. They’re not there for entertainment when they’re asking a question about how to respond to a new cancer diagnosis, for example, or what sort of retirement plan they should be subscribing to. So you don’t want content farms and random Reddit posts and garbage to be the results that are returned. You want to have reputable search results.

That framework of Your Money or Your Life has informed Google’s work on these high-stakes topics for quite some time. And that’s why I think it’s disturbing for people to see the AI-generated search results regurgitating clearly wrong health information from low-quality sites that perhaps happened to be in the training data.

So it seems like AI overviews is not following that same policy—or that’s what it appears like from the outside?

DiResta: That’s how it appears from the outside. I don’t know how they’re thinking about it internally. But those screenshots you’re seeing—a lot of these instances are being traced back to an isolated social media post or a clinic that’s disreputable but exists—are out there on the Internet. It’s not simply making things up. But it’s also not returning what we would consider to be a high-quality result in formulating its response.

I saw that Google responded to some of the problems with a blog post saying that it is aware of these poor results and it’s trying to make improvements. And I can read you the one bullet point that addressed health. It said, “For topics like news and health, we already have strong guardrails in place. In the case of health, we launched additional triggering refinements to enhance our quality protections.” Do you know what that means?

DiResta: That blog posts is an explanation that [AI Overviews] isn’t simply hallucinating—the fact that it’s pointing to URLs is supposed to be a guardrail because that enables the user to go and follow the result to its source. This is a good thing. They should be including those sources for transparency and so that outsiders can review them. However, it is also a fair bit of onus to put on the audience, given the trust that Google has built up over time by returning high-quality results in its health information search rankings.

I know one topic that you’ve tracked over the years has been disinformation about vaccine safety. Have you seen any evidence of that kind of disinformation making its way into AI search?

DiResta: I haven’t, though I imagine outside research teams are now testing results to see what appears. Vaccines have been so much a focus of the conversation around health misinformation for quite some time, I imagine that Google has had people looking specifically at that topic in internal reviews, whereas some of these other topics might be less in the forefront of the minds of the quality teams that are tasked with checking if there are bad results being returned.

What do you think Google’s next moves should be to prevent medical misinformation in AI search?

DiResta: Google has a perfectly good policy to pursue. Your Money or Your Life is a solid ethical guideline to incorporate into this manifestation of the future of search. So it’s not that I think there’s a new and novel ethical grounding that needs to happen. I think it’s more ensuring that the ethical grounding that exists remains foundational to the new AI search tools.

by Samuel K. Moore on 12. June 2024. at 15:00

For years, Nvidia has dominated many machine learning benchmarks, and now there are two more notches in its belt.

MLPerf, the AI benchmarking suite sometimes called “the Olympics of machine learning,” has released a new set of training tests to help make more and better apples-to-apples comparisons between competing computer systems. One of MLPerf’s new tests concerns fine-tuning of large language models, a process that takes an existing trained model and trains it a bit more with specialized knowledge to make it fit for a particular purpose. The other is for graph neural networks, a type of machine learning behind some literature databases, fraud detection in financial systems, and social networks.

Even with the additions and the participation of computers using Google’s and Intel’s AI accelerators, systems powered by Nvidia’s Hopper architecture dominated the results once again. One system that included 11,616 Nvidia H100 GPUs—the largest collection yet—topped each of the nine benchmarks, setting records in five of them (including the two new benchmarks).

“If you just throw hardware at the problem, it’s not a given that you’re going to improve.” —Dave Salvator, Nvidia

The 11,616-H100 system is “the biggest we’ve ever done,” says Dave Salvator, director of accelerated computing products at Nvidia. It smashed through the GPT-3 training trial in less than 3.5 minutes. A 512-GPU system, for comparison, took about 51 minutes. (Note that the GPT-3 task is not a full training, which could take weeks and cost millions of dollars. Instead, the computers train on a representative portion of the data, at an agreed-upon point well before completion.)

Compared to Nvidia’s largest entrant on GPT-3 last year, a 3,584 H100 computer, the 3.5-minute result represents a 3.2-fold improvement. You might expect that just from the difference in the size of these systems, but in AI computing that isn’t always the case, explains Salvator. “If you just throw hardware at the problem, it’s not a given that you’re going to improve,” he says.

“We are getting essentially linear scaling,” says Salvator. By that he means that twice as many GPUs lead to a halved training time. “[That] represents a great achievement from our engineering teams,” he adds.

Competitors are also getting closer to linear scaling. This round Intel deployed a system using 1,024 GPUs that performed the GPT-3 task in 67 minutes versus a computer one-fourth the size that took 224 minutes six months ago. Google’s largest GPT-3 entry used 12-times the number of TPU v5p accelerators as its smallest entry and performed its task nine times as fast.

Linear scaling is going to be particularly important for upcoming “AI factories” housing 100,000 GPUs or more, Salvator says. He says to expect one such data center to come online this year, and another, using Nvidia’s next architecture, Blackwell, to startup in 2025.

Nvidia’s streak continues

Nvidia continued to boost training times despite using the same architecture, Hopper, as it did in last year’s training results. That’s all down to software improvements, says Salvator. “Typically, we’ll get a 2-2.5x [boost] from software after a new architecture is released,” he says.

For GPT-3 training, Nvidia logged a 27 percent improvement from the June 2023 MLPerf benchmarks. Salvator says there were several software changes behind the boost. For example, Nvidia engineers tuned up Hopper’s use of less accurate, 8-bit floating point operations by trimming unnecessary conversions between 8-bit and 16-bit numbers and better targeting of which layers of a neural network could use the lower precision number format. They also found a more intelligent way to adjust the power budget of each chip’s compute engines, and sped communication among GPUs in a way that Salvator likened to “buttering your toast while it’s still in the toaster.”

Additionally, the company implemented a scheme called flash attention. Invented in the Stanford University laboratory of Samba Nova founder Chris Re, flash attention is an algorithm that speeds transformer networks by minimizing writes to memory. When it first showed up in MLPerf benchmarks, flash attention shaved as much as 10 percent from training times. (Intel, too, used a version of flash attention but not for GPT-3. It instead used the algorithm for one of the new benchmarks, fine-tuning.)

Using other software and network tricks, Nvidia delivered an 80 percent speedup in the text-to-image test, Stable Diffusion, versus its submission in November 2023.

New benchmarks

MLPerf adds new benchmarks and upgrades old ones to stay relevant to what’s happening in the AI industry. This year saw the addition of fine-tuning and graph neural networks.

Fine tuning takes an already trained LLM and specializes it for use in a particular field. Nvidia, for example took a trained 43-billion-parameter model and trained it on the GPU-maker’s design files and documentation to create ChipNeMo, an AI intended to boost the productivity of its chip designers. At the time, the company’s chief technology officer Bill Dally said that training an LLM was like giving it a liberal arts education, and fine tuning was like sending it to graduate school.

The MLPerf benchmark takes a pretrained Llama-2-70B model and asks the system to fine tune it using a dataset of government documents with the goal of generating more accurate document summaries.

There are several ways to do fine-tuning. MLPerf chose one called low-rank adaptation (LoRA). The method winds up training only a small portion of the LLM’s parameters leading to a 3-fold lower burden on hardware and reduced use of memory and storage versus other methods, according to the organization.

The other new benchmark involved a graph neural network (GNN). These are for problems that can be represented by a very large set of interconnected nodes, such as a social network or a recommender system. Compared to other AI tasks, GNNs require a lot of communication between nodes in a computer.

The benchmark trained a GNN on a database that shows relationships about academic authors, papers, and institutes—a graph with 547 million nodes and 5.8 billion edges. The neural network was then trained to predict the right label for each node in the graph.

Future fights

Training rounds in 2025 may see head-to-head contests comparing new accelerators from AMD, Intel, and Nvidia. AMD’s MI300 series was launched about six months ago, and a memory-boosted upgrade the MI325x is planned for the end of 2024, with the next generation MI350 slated for 2025. Intel says its Gaudi 3, generally available to computer makers later this year, will appear in MLPerf’s upcoming inferencing benchmarks. Intel executives have said the new chip has the capacity to beat H100 at training LLMs. But the victory may be short-lived, as Nvidia has unveiled a new architecture, Blackwell, which is planned for late this year.

by Dina Genkina on 12. June 2024. at 14:00

As large supercomputers keep getting larger, Sunnyvale, California-based Cerebras has been taking a different approach. Instead of connecting more and more GPUs together, the company has been squeezing as many processors as it can onto one giant wafer. The main advantage is in the interconnects—by wiring processors together on-chip, the wafer-scale chip bypasses many of the computational speed losses that come from many GPUs talking to each other, as well as losses from loading data to and from memory.

Now, Cerebras has flaunted the advantages of their wafer-scale chips in two separate but related results. First, the company demonstrated that its second generation wafer-scale engine, WSE-2, was significantly faster than world’s fastest supercomputer, Frontier, in molecular dynamics calculations—the field that underlies protein folding, modeling radiation damage in nuclear reactors, and other problems in material science. Second, in collaboration with machine learning model optimization company Neural Magic, Cerebras demonstrated that a sparse large language model could perform inference at one-third of the energy cost of a full model without losing any accuracy. Although the results are in vastly different fields, they were both possible because of the interconnects and fast memory access enabled by Cerebras’ hardware.

Speeding Through the Molecular World

“Imagine there’s a tailor and he can make a suit in a week,” says Cerebras CEO and co-founder Andrew Feldman. “He buys the neighboring tailor, and she can also make a suit in a week, but they can’t work together. Now, they can now make two suits in a week. But what they can’t do is make a suit in three and a half days.”

According to Feldman, GPUs are like tailors that can’t work together, at least when it comes to some problems in molecular dynamics. As you connect more and more GPUs, they can simulate more atoms at the same time, but they can’t simulate the same number of atoms more quickly.

Cerebras’ wafer-scale engine, however, scales in a fundamentally different way. Because the chips are not limited by interconnect bandwidth, they can communicate quickly, like two tailors collaborating perfectly to make a suit in three and a half days.

“It’s difficult to create materials that have the right properties, that have a long lifetime and sufficient strength and don’t break.” —Tomas Oppelstrup, Lawrence Livermore National Laboratory

To demonstrate this advantage, the team simulated 800,000 atoms interacting with each other, calculating the interactions in increments of one femtosecond at a time. Each step took just microseconds to compute on their hardware. Although that’s still 9 orders of magnitude slower than the actual interactions, it was also 179 times as fast as the Frontier supercomputer. The achievement effectively reduced a year’s worth of computation to just two days.

This work was done in collaboration with Sandia, Lawrence Livermore, and Los Alamos National Laboratories. Tomas Oppelstrup, staff scientist at Lawrence Livermore National Laboratory, says this advance makes it feasible to simulate molecular interactions that were previously inaccessible.

Oppelstrup says this will be particularly useful for understanding the longer-term stability of materials in extreme conditions. “When you build advanced machines that operate at high temperatures, like jet engines, nuclear reactors, or fusion reactors for energy production,” he says, “you need materials that can withstand these high temperatures and very harsh environments. It’s difficult to create materials that have the right properties, that have a long lifetime and sufficient strength and don’t break.” Being able to simulate the behavior of candidate materials for longer, Oppelstrup says, will be crucial to the material design and development process.

Ilya Sharapov, principal engineer at Cerebras, say the company is looking forward to extending applications of its wafer-scale engine to a larger class of problems, including molecular dynamics simulations of biological processes and simulations of airflow around cars or aircrafts.

Downsizing Large Language Models

As large language models (LLMs) are becoming more popular, the energy costs of using them are starting to overshadow the training costs—potentially by as much as a factor of ten in some estimates. “Inference is is the primary workload of AI today because everyone is using ChatGPT,” says James Wang, director of product marketing at Cerebras, “and it’s very expensive to run especially at scale.”

One way to reduce the energy cost (and speed) of inference is through sparsity—essentially, harnessing the power of zeros. LLMs are made up of huge numbers of parameters. The open-source Llama model used by Cerebras, for example, has 7 billion parameters. During inference, each of those parameters is used to crunch through the input data and spit out the output. If, however, a significant fraction of those parameters are zeros, they can be skipped during the calculation, saving both time and energy.

The problem is that skipping specific parameters is a difficult to do on a GPU. Reading from memory on a GPU is relatively slow, because they’re designed to read memory in chunks, which means taking in groups of parameters at a time. This doesn’t allow GPUs to skip zeros that are randomly interspersed in the parameter set. Cerebras CEO Feldman offered another analogy: “It’s equivalent to a shipper, only wanting to move stuff on pallets because they don’t want to examine each box. Memory bandwidth is the ability to examine each box to make sure it’s not empty. If it’s empty, set it aside and then not move it.”

“There’s a million cores in a very tight package, meaning that the cores have very low latency, high bandwidth interactions between them.” —Ilya Sharapov, Cerebras

Some GPUs are equipped for a particular kind of sparsity, called 2:4, where exactly two out of every four consecutively stored parameters are zeros. State-of-the-art GPUs have terabytes per second of memory bandwidth. The memory bandwidth of Cerebras’ WSE-2 is more than one thousand times as high, at 20 petabytes per second. This allows for harnessing unstructured sparsity, meaning the researchers can zero out parameters as needed, wherever in the model they happen to be, and check each one on the fly during a computation. “Our hardware is built right from day one to support unstructured sparsity,” Wang says.

Even with the appropriate hardware, zeroing out many of the model’s parameters results in a worse model. But the joint team from Neural Magic and Cerebras figured out a way to recover the full accuracy of the original model. After slashing 70 percent of the parameters to zero, the team performed two further phases of training to give the non-zero parameters a chance to compensate for the new zeros.

This extra training uses about 7 percent of the original training energy, and the companies found that they recover full model accuracy with this training. The smaller model takes one-third of the time and energy during inference as the original, full model. “What makes these novel applications possible in our hardware,” Sharapov says, “Is that there’s a million cores in a very tight package, meaning that the cores have very low latency, high bandwidth interactions between them.”

by John Boyd on 10. June 2024. at 13:00

As Intel, Samsung, TSMC, and Japan’s upcoming advanced foundry Rapidus each make their separate preparations to cram more and more transistors into every square millimeter of silicon, one thing they all have in common is that the extreme ultraviolet (EUV) lithography technology underpinning their efforts is extremely complex, extremely expensive, and extremely costly to operate. A prime reason is that the source of this system’s 13.5-nanometer light is the precise and costly process of blasting flying droplets of molten tin with the most powerful commercial lasers on the planet.

But an unconventional alternative is in the works. A group of researchers at the High Energy Accelerator Research Organization, known as KEK, in Tsukuba, Japan, is betting EUV lithography might be cheaper, quicker, and more efficient if it harnesses the power of a particle accelerator.

Even before the first EUV machines had been installed in fabs, researchers saw possibilities for EUV lithography using a powerful light source called a free-electron laser (FEL), which is generated by a particle accelerator. However, not just any particle accelerator will do, say the scientists at KEK. They claim the best candidate for EUV lithography incorporates the particle-accelerator version of regenerative braking. Known as an energy recovery linear accelerator, it could enable a free electron laser to economically generate tens of kilowatts of EUV power. This is more than enough to drive not one but many next-generation lithography machines simultaneously, pushing down the cost of advanced chipmaking.

“The FEL beam’s extreme power, its narrow spectral width, and other features make it suitable as an application for future lithography,” Norio Nakamura, researcher in advanced light sources at KEK, told me on a visit to the facility.

Linacs Vs. Laser-Produced Plasma

Today’s EUV systems are made by a single manufacturer, ASML, headquartered in Veldhoven, Netherlands. When ASML introduced the first generation of these US $100-million-plus precision machines in 2016, the industry was desperate for them. Chipmakers had been getting by with workaround after workaround for the then most advanced system, lithography using 193-nm light. Moving to a much shorter, 13.5-nm wavelength was a revolution that would collapse the number of steps needed in chipmaking and allow Moore’s Law to continue well into the next decade.

The chief cause of the continual delays was a light source that was too dim. The technology that ultimately delivered a bright enough source of EUV light is called laser-produced plasma, or EUV-LPP. It employs a carbon dioxide laser to blast molten droplets of tin into plasma thousands of times per second. The plasma emits a spectrum of photonic energy, and specialized optics then capture the necessary 13.5-nm wavelength from the spectrum and guide it through a sequence of mirrors. Subsequently, the EUV light is reflected off a patterned mask and then projected onto a silicon wafer.

The experimental compact energy recovery linac at KEK uses most of the energy from electrons on a return journey to speed up a new set of electrons.KEK

It all adds up to a highly complex process. And although it starts off with kilowatt-consuming lasers, the amount of EUV light that is reflected onto the wafer is just several watts. The dimmer the light, the longer it takes to reliably expose a pattern on the silicon. Without enough photons carrying the pattern, EUV would be uneconomically slow. And pushing too hard for speed can lead to costly errors.

When the machines were first introduced, the power level was enough to process about 100 wafers per hour. Since then, ASML has managed to steadily hike the output to about 200 wafers per hour for the present series of machines.

ASML’s current light sources are rated at 500 watts. But for the even finer patterning needed in the future, Nakamura says it could take 1 kilowatt or more. ASML says it has a road map to develop a 1,000-W light source. But it could be difficult to achieve, says Nakamura, who formerly led the beam dynamics and magnet group at KEK and came out of retirement to work on the EUV project.

Difficult but not necessarily impossible. Doubling the source power is “very challenging,” agrees Ahmed Hassanein who leads the Center for Materials Under Extreme Environment, at Purdue University, in Indiana. But he points out that ASML has achieved similarly difficult targets in the past using an integrated approach of improving and optimizing the light source and other components, and he isn’t ruling out a repeat.

In a free electron laser, accelerated electrons are subject to alternating magnetic fields, causing them to undulate and emit electromagnetic radiation. The radiation bunches up the electrons, leading to their amplifying only a specific wavelength, creating a laser beam.Chris Philpot

But brightness isn’t the only issue ASML faces with laser-produced plasma sources. “There are a number of challenging issues in upgrading to higher EUV power,” says Hassanein. He rattles off several, including “contamination, wavelength purity, and the performance of the mirror-collection system.”

High operating costs are another problem. These systems consume some 600 liters of hydrogen gas per minute, most of which goes into keeping tin and other contaminants from getting onto the optics and wafers. (Recycling, however, could reduce this figure.)

But ultimately, operating costs come down to electricity consumption. Stephen Benson, recently retired senior research scientist at the Thomas Jefferson National Accelerator Facility, in Virginia., estimates that the wall-plug efficiency of the whole EUV-LPP system might be less than 0.1 percent. Free electron lasers, like the one KEK is developing, could be as much as 10 to 100 times as efficient, he says.

The Energy Recovery Linear Accelerator

The system KEK is developing generates light by boosting electrons to relativistic speeds and then deviating their motion in a particular way.

The process starts, Nakamura explains, when an electron gun injects a beam of electrons into a meters-long cryogenically cooled tube. Inside this tube, superconductors deliver radio-frequency (RF) signals that drive the electrons along faster and faster. The electrons then make a 180-degree turn and enter a structure called an undulator, a series of oppositely oriented magnets. (The KEK system currently has two.) The undulators force the speeding electrons to follow a sinusoidal path, and this motion causes the electrons to emit light.

In linear accelerator, injected electrons gain energy from an RF field. Ordinarily, the electrons would then enter a free electron laser and are immediately disposed of in a beam dump. But in an energy recovery linear accelerator (ERL), the electrons circle back into the RF field and lend their energy to newly injected electrons before exiting to a beam dump.

What happens next is a phenomenon called self-amplified spontaneous emissions, or SASE. The light interacts with the electrons, slowing some and speeding up others, so they gather into “microbunches,” peaks in density that occur periodically along the undulator’s path. The now-structured electron beam amplifies only the light that’s in phase with the period of these microbunches, generating a coherent beam of laser light.

It’s at this point that KEK’s compact energy recovery linac (cERL), diverges from lasers driven by conventional linear accelerators. Ordinarily, the spent beam of electrons is disposed of by diverting the particles into what is called a beam dump. But in the cERL, the electrons first loop back into the RF accelerator. This beam is now in the opposite phase to newly injected electrons that are just starting their journey. The result is that the spent electrons transfer much of their energy to the new beam, boosting its energy. Once the original electrons have had some of their energy drained away like this, they are diverted into a beam dump.

“The acceleration energy in the linac is recovered, and the dumped beam power is drastically reduced compared to [that of] an ordinary linac,” Nakamura explains to me while scientists in another room operate the laser. Reusing the electrons’ energy means that for the same amount of electricity the system sends more current through the accelerator and can fire the laser more frequently, he says.

Other experts agree. The energy-recovery linear accelerator’s improved efficiency can lower costs, “which is a major concern of using EUV laser-produced plasma,” says Hassanein.

The Energy Recovery Linac for EUV

The KEK compact energy-recovery linear accelerator was initially constructed between 2011 and 2013 with the aim of demonstrating its potential as a synchrotron radiation source for researchers working for the institution’s physics and materials-science divisions. But researchers were dissatisfied with the planned system, which had a lower performance target than could be achieved by some storage ring-based synchrotrons—huge circular accelerators that keep a beam of electrons moving with a constant kinetic energy. So, the KEK researchers went in search of a more appropriate application. After talking with Japanese tech companies, including Toshiba, which had a flash memory chip division at the time, the researchers conducted an initial study that confirmed that a kilowatt-class light source was possible with a compact energy-recovery linear accelerator. And so, the EUV free-electron-laser project was born. In 2019 and 2020, the researchers modified the existing experimental accelerator to start the journey to EUV light.

The system is housed in an all-concrete room to protect researchers from the intense electromagnetic radiation produced during operation. The room is some 60 meters long and 20 meters wide with much of the space taken up by a bewildering tangle of complex equipment, pipes, and cables that snakes along both sides of its length in the form of an elongated racetrack.

The accelerator is not yet able to generate EUV wavelengths. With an electron beam energy of 17 megaelectronvolts, the researchers have been able to generate SASE emissions in bursts of 20-micrometer infrared light. Early test results were published in the Japanese Journal of Applied Physics in April 2023. The next step, which is underway, is to generate much greater laser power in continuous-wave mode.

To be sure, 20 micrometers is a far cry from 13.5 nanometers. And there are already types of particle accelerators that produce synchrotron radiation of even shorter wavelengths than EUV. But lasers based on energy-recovery linear accelerators could generate significantly more EUV power due to their inherent efficiency, the KEK researchers claim. In synchrotron radiation sources, light intensity increases proportionally to the number of injected electrons. By comparison, in free-electron laser systems, light intensity increases roughly with the square of the number of injected electrons, resulting in much more brightness and power.

For an energy-recovery linear accelerator to reach the EUV range will require equipment upgrades beyond what KEK currently has room for. So, the researchers are now making the case for constructing a new prototype system that can produce the needed 800 MeV.

An electron gun injects charge into the compact energy recovery linear accelerator at KEK.KEK

In 2021, before severe inflation affected economies around the globe, the KEK team estimated the construction cost (excluding land) for a new system at 40 billion yen ($260 million) for a system that delivers 10 kW of EUV and supplies multiple lithography machines. Annual running costs were judged to be about 4 billion yen. So even taking recent inflation into account, “the estimated costs per exposure tool in our setup are still rather low compared to the estimated costs” for today’s laser-produced plasma source, says Nakamura.

There are plenty of technical challenges to work out before such a system can achieve the high levels of performance and stability of operations demanded by semiconductor manufacturers, admits Nakamura. The team will have to develop new editions of key components such as the superconducting cavity, the electron gun, and the undulator. Engineers will also have to develop good procedural techniques to ensure, for instance, that the electron beam does not degrade or falter during operations.

And to ensure their approach is cost effective enough to grab the attention of chipmakers, the researchers will need to create a system that can reliably transport more than 1 kW of EUV power simultaneously to multiple lithography machines. The researchers already have a conceptual design for an arrangement of special mirrors that would convey the EUV light to multiple exposure tools without significant loss of power or damage to the mirrors.

Other EUV Possibilities

It’s too early in the development of EUV free-electron lasers for rapidly expanding chipmakers to pay it much attention. But the KEK team is not alone in chasing the technology. A venture-backed startup xLight, in Palo Alto, Calif. is also among those chasing it. The company, which is packed with particle-accelerator veterans from the Stanford Linear Accelerator and elsewhere, recently inked an R&D deal with Fermi National Accelerator Laboratory, in Illinois, to develop superconducting cavities and cryomodule technology. Attempts to contact xLight went unanswered, but in January, the company took part in the 8th Workshop EUV-FEL in Tokyo, and former CEO Erik Hosler gave a presentation on the technology.

Significantly, ASML considered turning to particle accelerators a decade ago and again more recently when it compared the progress of free-electron laser technology to the laser-produced plasma road map. But company executives decided LLP presented fewer risks.

And, indeed, it is a risky road. Independent views on KEK’s project emphasize that reliability and funding will be the biggest challenges the researchers face going forward. “The R&D road map will involve numerous demanding stages in order to develop a reliable, mature system,” says Hassanein. “This will require serious investment and take considerable time.”

“The machine design must be extremely robust, with redundancy built in,” adds retired research scientist Benson. The design must also ensure that components are not damaged from radiation or laser light.” And this must be accomplished “without compromising performance, which must be good enough to ensure decent wall-plug efficiency.”

More importantly, Benson warns that without a forthcoming commitment to invest in the technology, “development of EUV-FELs might not come in time to help the semiconductor industry.”

by Benjie Holson on 09. June 2024. at 13:00

The original version of this post by Benjie Holson was published on Substack here, and includes Benjie’s original comics as part of his series on robots and startups.

I worked on this idea for months before I decided it was a mistake. The second time I heard someone mention it, I thought, “That’s strange, these two groups had the same idea. Maybe I should tell them it didn’t work for us.” The third and fourth time I rolled my eyes and ignored it. The fifth time I heard about a group struggling with this mistake, I decided it was worth a blog post all on its own. I call this idea “The Mythical Non-Roboticist.”

---

The Mistake

The idea goes something like this: Programming robots is hard. And there are some people with really arcane skills and PhDs who are really expensive and seem to be required for some reason. Wouldn’t it be nice if we could do robotics without them? ¹ What if everyone could do robotics? That would be great, right? We should make a software framework so that non-roboticists can program robots.

This idea is so close to a correct idea that it’s hard to tell why it doesn’t work out. On the surface, it’s not wrong: All else being equal, it would be good if programming robots was more accessible. The problem is that we don’t have a good recipe for making working robots. So we don’t know how to make that recipe easier to follow. In order to make things simple, people end up removing things that folks might need, because no one knows for sure what’s absolutely required. It’s like saying you want to invent an invisibility cloak and want to be able to make it from materials you can buy from Home Depot. Sure, that would be nice, but if you invented an invisibility cloak that required some mercury and neodymium to manufacture would you toss the recipe?

In robotics, this mistake is based on a very true and very real observation: Programming robots is super hard. Famously hard. It would be super great if programming robots was easier. The issue is this: Programming robots has two different kinds of hard parts.

Robots are hard because the world is complicated

Moor Studio/Getty Images

The first kind of hard part is that robots deal with the real world, imperfectly sensed and imperfectly actuated. Global mutable state is bad programming style because it’s really hard to deal with, but to robot software the entire physical world is global mutable state, and you only get to unreliably observe it and hope your actions approximate what you wanted to achieve. Getting robotics to work at all is often at the very limit of what a person can reason about, and requires the flexibility to employ whatever heuristic might work for your special problem. This is the intrinsic complexity of the problem: Robots live in complex worlds, and for every working solution there are millions of solutions that don’t work, and finding the right one is hard, and often very dependent on the task, robot, sensors, and environment.

Folks look at that challenge, see that it is super hard, and decide that, sure, maybe some fancy roboticist could solve it in one particular scenario, but what about “normal” people? “We should make this possible for non-roboticists” they say. I call these users “Mythical Non-Roboticists” because once they are programming a robot, I feel they become roboticists. Isn’t anyone programming a robot for a purpose a roboticist? Stop gatekeeping, people.

Don’t design for amorphous groups

I call also them “mythical” because usually the “non-roboticist” implied is a vague, amorphous group. Don’t design for amorphous groups. If you can’t name three real people (that you have talked to) that your API is for, then you are designing for an amorphous group and only amorphous people will like your API.

And with this hazy group of users in mind (and seeing how difficult everything is), folks think, “Surely we could make this easier for everyone else by papering over these things with simple APIs?”

No. No you can’t. Stop it.

You can’t paper over intrinsic complexity with simple APIs because if your APIs are simple they can’t cover the complexity of the problem. You will inevitably end up with a beautiful looking API, with calls like “grasp_object” and “approach_person” which demo nicely in a hackathon kickoff but last about 15 minutes of someone actually trying to get some work done. It will turn out that, for their particular application, “grasp_object()” makes 3 or 4 wrong assumptions about “grasp” and “object” and doesn’t work for them at all.

Your users are just as smart as you

This is made worse by the pervasive assumption that these people are less savvy (read: less intelligent) than the creators of this magical framework. ² That feeling of superiority will cause the designers to cling desperately to their beautiful, simple “grasp_object()”s and resist adding the knobs and arguments needed to cover more use cases and allow the users to customize what they get.

Ironically this foists a bunch of complexity on to the poor users of the API who have to come up with clever workarounds to get it to work at all.

Moor Studio/Getty Images

The sad, salty, bitter icing on this cake-of-frustration is that, even if done really well, the goal of this kind of framework would be to expand the group of people who can do the work. And to achieve that, it would sacrifice some performance you can only get by super-specializing your solution to your problem. If we lived in a world where expert roboticists could program robots that worked really well, but there was so much demand for robots that there just wasn’t enough time for those folks to do all the programming, this would be a great solution. ³

The obvious truth is that (outside of really constrained environments like manufacturing cells) even the very best collection of real bone-fide, card-carrying roboticists working at the best of their ability struggle to get close to a level of performance that makes the robots commercially viable, even with long timelines and mountains of funding. ⁴ We don’t have any headroom to sacrifice power and effectiveness for ease.

What problem are we solving?

So should we give up making it easier? Is robotic development available only to a small group of elites with fancy PhDs? ⁵ No to both! I have worked with tons of undergrad interns who have been completely able to do robotics.⁶ I myself am mostly self-taught in robot programming.⁷ While there is a lot of intrinsic complexity in making robots work, I don’t think there is any more than, say, video game development.

In robotics, like in all things, experience helps, some things are teachable, and as you master many areas you can see things start to connect together. These skills are not magical or unique to robotics. We are not as special as we like to think we are.

But what about making programming robots easier? Remember way back at the beginning of the post when I said that there were two different kinds of hard parts? One is the intrinsic complexity of the problem, and that one will be hard no matter what. ⁸ But the second is the incidental complexity, or as I like to call it, the stupid BS complexity.

Stupid BS Complexity

Robots are asynchronous, distributed, real-time systems with weird hardware. All of that will be hard to configure for stupid BS reasons. Those drivers need to work in the weird flavor of Linux you want for hard real-time for your controls and getting that all set up will be hard for stupid BS reasons. You are abusing Wi-Fi so you can roam seamlessly without interruption but Linux’s Wi-Fi will not want to do that. Your log files are huge and you have to upload them somewhere so they don’t fill up your robot. You’ll need to integrate with some cloud something or other and deal with its stupid BS. ⁹

Moor Studio/Getty Images

There is a ton of crap to deal with before you even get to complexity of dealing with 3D rotation, moving reference frames, time synchronization, messaging protocols. Those things have intrinsic complexity (you have to think about when something was observed and how to reason about it as other things have moved) and stupid BS complexity (There’s a weird bug because someone multiplied two transform matrices in the wrong order and now you’re getting an error message that deep in some protocol a quaternion is not normalized. WTF does that mean?) ¹⁰

One of the biggest challenges of robot programming is wading through the sea of stupid BS you need to wrangle in order to start working on your interesting and challenging robotics problem.

So a simple heuristic to make good APIs is:

Design your APIs for someone as smart as you, but less tolerant of stupid BS.

That feels universal enough that I’m tempted to call it Holson’s Law of Tolerable API Design.

When you are using tools you’ve made, you know them well enough to know the rough edges and how to avoid them.

But rough edges are things that have to be held in a programmer’s memory while they are using your system. If you insist on making a robotics framework ¹¹, you should strive to make it as powerful as you can with the least amount of stupid BS. Eradicate incidental complexity everywhere you can. You want to make APIs that have maximum flexibility but good defaults. I like python’s default-argument syntax for this because it means you can write APIs that can be used like:

It is possible to have easy things be simple and allow complex things. And please, please, please don’t make condescending APIs. Thanks!

1. Ironically it is very often the expensive arcane-knowledge-having PhDs who are proposing this.

2. Why is it always a framework?

3. The exception that might prove the rule is things like traditional manufacturing-cell automation. That is a place where the solutions exist, but the limit to expanding is set up cost. I’m not an expert in this domain, but I’d worry that physical installation and safety compliance might still dwarf the software programming cost, though.

4. As I well know from personal experience.

5. Or non-fancy PhDs for that matter?

6. I suspect that many bright highschoolers would also be able to do the work. Though, as Google tends not to hire them, I don’t have good examples.

7. My schooling was in Mechanical Engineering and I never got a PhD, though my ME classwork did include some programming fundamentals.

8. Unless we create effective general purpose AI. It feels weird that I have to add that caveat, but the possibility that it’s actually coming for robotics in my lifetime feels much more possible than it did two years ago.

9. And if you are unlucky, its API was designed by someone who thought they were smarter than their customers.

10. This particular flavor of BS complexity is why I wrote posetree.py. If you do robotics, you should check it out.

11. Which, judging by the trail of dead robot-framework-companies, is a fraught thing to do.

by Evan Ackerman on 07. June 2024. at 16:00

Video Friday is your weekly selection of awesome robotics videos, collected by your friends at IEEE Spectrum robotics. We also post a weekly calendar of upcoming robotics events for the next few months. Please send us your events for inclusion.

RoboCup 2024: 17–22 July 2024, EINDHOVEN, NETHERLANDS

ICRA@40: 23–26 September 2024, ROTTERDAM, NETHERLANDS

IROS 2024: 14–18 October 2024, ABU DHABI, UNITED ARAB EMIRATES

ICSR 2024: 23–26 October 2024, ODENSE, DENMARK

Cybathlon 2024: 25–27 October 2024, ZURICH

Enjoy today’s videos!

In this video, you see the start of 1X’s development of an advanced AI system that chains simple tasks into complex actions using voice commands, allowing seamless multi-robot control and remote operation. By starting with single-task models, we ensure smooth transitions to more powerful unified models, ultimately aiming to automate high-level actions using AI.

This video does not contain teleoperation, computer graphics, cuts, video speedups, or scripted trajectory playback. It’s all controlled via neural networks.

[ 1X ]

As the old adage goes, one cannot claim to be a true man without a visit to the Great Wall of China. XBot-L, a full-sized humanoid robot developed by Robot Era, recently acquitted itself well in a walk along sections of the Great Wall.

[ Robot Era ]

The paper presents a novel rotary wing platform, that is capable of folding and expanding its wings during flight. Our source of inspiration came from birds’ ability to fold their wings to navigate through small spaces and dive. The design of the rotorcraft is based on the monocopter platform, which is inspired by the flight of Samara seeds.

[ AirLab ]

We present a variable stiffness robotic skin (VSRS), a concept that integrates stiffness-changing capabilities, sensing, and actuation into a single, thin modular robot design. Reconfiguring, reconnecting, and reshaping VSRSs allows them to achieve new functions both on and in the absence of a host body.

[ Yale Faboratory ]

Heimdall is a new rover design for the 2024 University Rover Challenge (URC). This video shows highlights of Heimdall’s trip during the four missions at URC 2024.

Heimdall features a split body design with whegs (wheel legs), and a drill for sub-surface sample collection. It also has the ability to manipulate a variety of objects, collect surface samples, and perform onboard spectrometry and chemical tests.

[ WVU ]

I think this may be the first time I’ve seen an autonomous robot using a train? This one is delivering lunch boxes!

[ JSME ]

The AI system used identifies and separates red apples from green apples, after which a robotic arm picks up the red apples identified with a qb SoftHand Industry and gently places them in a basket.

My favorite part is the magnetic apple stem system.

[ QB Robotics ]

DexNex (v0, June 2024) is an anthropomorphic teleoperation testbed for dexterous manipulation at the Center for Robotics and Biosystems at Northwestern University. DexNex recreates human upper-limb functionality through a near 1-to-1 mapping between Operator movements and Avatar actions.

Motion of the Operator’s arms, hands, fingers, and head are fed forward to the Avatar, while fingertip pressures, finger forces, and camera images are fed back to the Operator. DexNex aims to minimize the latency of each subsystem to provide a seamless, immersive, and responsive user experience. Future research includes gaining a better understanding of the criticality of haptic and vision feedback for different manipulation tasks; providing arm-level grounded force feedback; and using machine learning to transfer dexterous skills from the human to the robot.

[ Northwestern ]

Sometimes the best path isn’t the smoothest or straightest surface, it’s the path that’s actually meant to be a path.

[ RaiLab ]

Fulfilling a school requirement by working in a Romanian locomotive factory one week each month, Daniela Rus learned to operate “machines that help us make things.” Appreciation for the practical side of math and science stuck with Daniela, who is now Director of the MIT Computer Science and Artificial Intelligence Laboratory (CSAIL).

[ MIT ]

For AI to achieve its full potential, non-experts need to be let into the development process, says Rumman Chowdhury, CEO and cofounder of Humane Intelligence. She tells the story of farmers fighting for the right to repair their own AI-powered tractors (which some manufacturers actually made illegal), proposing everyone should have the ability to report issues, patch updates or even retrain AI technologies for their specific uses.

[ TED ]

by Rahul Rao on 07. June 2024. at 11:00

Just as early mariners used simple compasses to chart courses across the sea, today’s ships, planes, satellites, and smartphones can rely on Earth’s magnetic field to find their bearings. The difference is that today’s rather more sophisticated compasses have the aid of complex models, like the commonly used World Magnetic Model (WMM), that try to capture the multifaceted processes that create Earth’s magnetosphere. A compass can rely on the WMM or similar models to convert a needle pointing to magnetic north to a heading with respect to true north. (The two norths differ by ever-changing angles.)

These models are not perfect: There are differences between the magnetosphere that they predict and the magnetosphere that satellites observe. Scientists have traditionally ascribed these differences to space currents that flow through the magnetic field high in Earth’s upper atmosphere. But new research complicates the picture, suggesting that the differences are the result of observational biases, incomplete models, or both.

For craft that require sensitive navigation, particularly around Earth’s poles, any of these complications pose a problem. And those problems stand to grow as polar ice melts around the North Pole, opening up potential new shipping routes.

Earth’s magnetic field is multifaceted and complex, but models like the WMM can project it out a few years at a time. The WMM’s current edition, released in December 2019, contains estimates of Earth’s magnetic field from the start of 2020 to the end of 2024. (The next version, covering 2025 through 2029, is scheduled for release in December of this year.)

“Compasses need to account for space currents already, but this adds more complication and sources of noise that have to be dealt with.” —Mark Moldwin, University of Michigan

These models do not always account for space currents, which are often pushed around by extraterrestrial forces like the solar wind. But if space currents are responsible for the discrepancies between models and observations, scientists could identify them by simply finding the differences, which they call “residuals.” Moreover, there would then be little reason for one of Earth’s hemispheres to display more residuals than the other—except that’s what existing models predict.

But the new study’s authors, space physicists Yining Shi and Mark Moldwin from the University of Michigan, had been among a number of researchers who had spotted an imbalance in residuals. More residuals seemed to emerge from the magnetic woodwork, so to speak, in the southern hemisphere than in the Northern Hemisphere. “We wanted to take a closer look at them,” Moldwin said.

Shi and Moldwin compared estimates between 2014 and 2020 from another Earth magnetic field model, IGRF-13, with observations from the European Space Agency’s Swarm mission, a trio of satellites that have continually measured Earth’s magnetic field since their 2014 launch.

When they focused on residuals over that time period, they did indeed find about 12 percent more major residuals in the Southern Hemisphere than in the Northern. All of these large residuals were found in the polar regions. Many were concentrated at latitudes of 70 degrees north and south, where scientists expect to find space currents.

But another spate of residuals were concentrated closer to Earth’s geographic poles, about 80 degrees north and south, where they have no obvious geophysical explanation. Moreover, the distributions of these poles differed—matching the fact that Earth’s geographic poles map to different magnetic coordinates.

This second peak in particular led the researchers to consider alternative explanations. It is possible, for instance, that IGRF-13 simply does not capture all of the factors driving Earth’s magnetosphere around the poles. But another cause could be the satellites themselves. Shi and Moldwin say that, because Swarm satellites reside in orbits that cross the poles, Earth’s northern and southern polar regions are overrepresented in their magnetic measurements.

“Compasses need to account for space currents already, but this adds more complication and sources of noise that have to be dealt with,” Moldwin said.

Now, Shi is examining these residuals more closely to pick apart the causes of the residuals—which ones have actual geophysical explanations and which are the result of statistical errors.

Shi and Moldwin published their work on 6 May in Journal of Geophysical Research: Space Physics.

by Tim Hornyak on 06. June 2024. at 14:51

Skies over Tokyo are thick with air traffic these days amid an influx of international tourists. But one plane recently helped revive the dream of airborne Internet access for all. Researchers in Japan announced on 28 May that they have successfully tested 5G communications equipment in the 38 gigahertz band from an altitude of 4 kilometers.

The experiment was aimed at developing an aerial relay backhaul with millimeter-wave band links between ground stations and a simulated High-Altitude Platform Station (HAPS), a radio station aboard an uncrewed aircraft that stays aloft in the stratosphere for extended periods of time. A Cessna flying out of Chofu Airfield in western Tokyo was outfitted with a 38 GHz 5G base station and core network device, and three ground stations were equipped with lens antennas with automatic tracking.

With the Cessna as a relay station, the setup enabled communication between one ground station connected to the 5G terrestrial network and a terrestrial base station connected to a user terminal, according to a consortium of Japanese companies and the National Institute of Information and Communications Technology.

“We developed technology that enables communication using 5G [New Radio] by correctly directing 38 GHz beams toward three ground stations while adapting to the flight attitude, speed, direction, position, altitude, etc. during aircraft rotation,” said Shinichi Tanaka, a manager in broadcaster SKY Perfect JSAT’s Space Business Division. “We confirmed that the onboard system, designed for the stratosphere, has adequate communication and tracking performance even under the flight speed and attitude fluctuations of a Cessna aircraft, which are more severe than those of HAPS.”

The sharpest beam width of the ground station antenna is 0.8 degrees, and the trial demonstrated a tracking method that always captures the Cessna in this angular range, Tanaka added.

A Cessna [top left] carried a 38 GHz antenna [top right] during a flight, functioning as a 5G base station for receivers on the ground [bottom right]. The plane was able to connect to multiple ground stations at once [illustration, bottom left].NTT Docomo

Millimeter wave bands, such as the 38 GHz band, have the highest data capacity for 5G and are suited for crowded venues such as stadiums and shopping centers. When used outdoors, however, the signals can be attenuated by rain and other moisture in the atmosphere. To counter this, the consortium successfully tested an algorithm that automatically switches between multiple ground stations to compensate for moisture-weakened signals.

Unlike Google’s failed Loon effort, which focused on providing direct communication to user terminals, the HAPS trial is aimed at creating backhaul lines for base stations. Led by Japan’s Ministry of Internal Affairs and Communications, the experiment is designed to deliver high-speed, high-capacity communications both for the development of 5G and 6G networks as well as emergency response. The latter is critical in disaster-prone Japan—in January, communication lines around the Noto Peninsula on the Sea of Japan were severed following a magnitude-7 earthquake that caused over 1,500 casualties.

“This is the world’s first successful 5G communication experiment via the sky using the Q-band frequency,” said Hinata Kohara, a researcher with mobile carrier NTT Docomo’s 6G Network Innovation Department. “In addition, the use of 5G communication base stations and core network equipment on the aircraft for communication among multiple ground stations enables flexible and fast route switching of the ground [gateway] station for a feeder link, and is robust against propagation characteristics such as rainfall. Another key feature is the use of a full digital beamforming method for beam control, which uses multiple independent beams to improve frequency utilization efficiency.”

Doppler shift compensation was a challenge in the experiment, Kohara said, adding that the researchers will conduct further tests to find a solution with the aim of commercializing a HAPS service in 2026. Aside from SKY Perfect JSAT and NTT Docomo, the consortium includes Panasonic Holdings, known for its electronics equipment.

The HAPS push comes as NTT Docomo announced it has led another consortium in a $100 million investment in Airbus’ AALTO HAPS, operator of the Zephyr fixed-wing uncrewed aerial vehicle. The solar-powered wing can be used for 5G direct-to-device communications or Earth observation, and has set records including 64 days of stratospheric flight. According to Airbus, it has a reach of “up to 250 terrestrial towers in difficult mountainous terrain.” Docomo said the investment is aimed at commercializing Zephyr services in Japan, including coverage of rural areas and disaster zones, and around the world in 2026.

by Glenn Zorpette on 06. June 2024. at 13:59

Generative artificial intelligence, and large language models in particular, are starting to change how countless technical and creative professionals do their jobs. Programmers, for example, are getting code segments by prompting large language models. And graphic arts software packages such as Adobe Illustrator already have tools built in that let designers conjure illustrations, images, or patterns by describing them.

But such conveniences barely hint at the massive, sweeping changes to employment predicted by some analysts. And already, in ways large and small, striking and subtle, the tech world’s notables are grappling with changes, both real and envisioned, wrought by the onset of generative AI. To get a better idea of how some of them view the future of generative AI, IEEE Spectrum asked three luminaries—an academic leader, a regulator, and a semiconductor industry executive—about how generative AI has begun affecting their work. The three, Andrea Goldsmith, Juraj Čorba, and Samuel Naffziger, agreed to speak with Spectrum at the 2024 IEEE VIC Summit & Honors Ceremony Gala, held in May in Boston.

Click to read more thoughts from:

1. Andrea Goldsmith, dean of engineering at Princeton University.
2. Juraj Čorba, senior expert on digital regulation and governance, Slovak Ministry of Investments, Regional Development
3. Samuel Naffziger, senior vice president and a corporate fellow at Advanced Micro Devices

Andrea Goldsmith

Andrea Goldsmith is dean of engineering at Princeton University.

There must be tremendous pressure now to throw a lot of resources into large language models. How do you deal with that pressure? How do you navigate this transition to this new phase of AI?

Andrea J. Goldsmith

Andrea Goldsmith: Universities generally are going to be very challenged, especially universities that don’t have the resources of a place like Princeton or MIT or Stanford or the other Ivy League schools. In order to do research on large language models, you need brilliant people, which all universities have. But you also need compute power and you need data. And the compute power is expensive, and the data generally sits in these large companies, not within universities.

So I think universities need to be more creative. We at Princeton have invested a lot of money in the computational resources for our researchers to be able to do—well, not large language models, because you can’t afford it. To do a large language model… look at OpenAI or Google or Meta. They’re spending hundreds of millions of dollars on compute power, if not more. Universities can’t do that.

But we can be more nimble and creative. What can we do with language models, maybe not large language models but with smaller language models, to advance the state of the art in different domains? Maybe it’s vertical domains of using, for example, large language models for better prognosis of disease, or for prediction of cellular channel changes, or in materials science to decide what’s the best path to pursue a particular new material that you want to innovate on. So universities need to figure out how to take the resources that we have to innovate using AI technology.

We also need to think about new models. And the government can also play a role here. The [U.S.] government has this new initiative, NAIRR, or National Artificial Intelligence Research Resource, where they’re going to put up compute power and data and experts for educators to use—researchers and educators.

That could be a game-changer because it’s not just each university investing their own resources or faculty having to write grants, which are never going to pay for the compute power they need. It’s the government pulling together resources and making them available to academic researchers. So it’s an exciting time, where we need to think differently about research—meaning universities need to think differently. Companies need to think differently about how to bring in academic researchers, how to open up their compute resources and their data for us to innovate on.

As a dean, you are in a unique position to see which technical areas are really hot, attracting a lot of funding and attention. But how much ability do you have to steer a department and its researchers into specific areas? Of course, I’m thinking about large language models and generative AI. Is deciding on a new area of emphasis or a new initiative a collaborative process?

Goldsmith: Absolutely. I think any academic leader who thinks that their role is to steer their faculty in a particular direction does not have the right perspective on leadership. I describe academic leadership as really about the success of the faculty and students that you’re leading. And when I did my strategic planning for Princeton Engineering in the fall of 2020, everything was shut down. It was the middle of COVID, but I’m an optimist. So I said, “Okay, this isn’t how I expected to start as dean of engineering at Princeton.” But the opportunity to lead engineering in a great liberal arts university that has aspirations to increase the impact of engineering hasn’t changed. So I met with every single faculty member in the School of Engineering, all 150 of them, one-on-one over Zoom.

And the question I asked was, “What do you aspire to? What should we collectively aspire to?” And I took those 150 responses, and I asked all the leaders and the departments and the centers and the institutes, because there already were some initiatives in robotics and bioengineering and in smart cities. And I said, “I want all of you to come up with your own strategic plans. What do you aspire to in these areas? And then let’s get together and create a strategic plan for the School of Engineering.” So that’s what we did. And everything that we’ve accomplished in the last four years that I’ve been dean came out of those discussions, and what it was the faculty and the faculty leaders in the school aspired to.

So we launched a bioengineering institute last summer. We just launched Princeton Robotics. We’ve launched some things that weren’t in the strategic plan that bubbled up. We launched a center on blockchain technology and its societal implications. We have a quantum initiative. We have an AI initiative using this powerful tool of AI for engineering innovation, not just around large language models, but it’s a tool—how do we use it to advance innovation and engineering? All of these things came from the faculty because, to be a successful academic leader, you have to realize that everything comes from the faculty and the students. You have to harness their enthusiasm, their aspirations, their vision to create a collective vision.

Juraj Čorba

Juraj Čorba is senior expert on digital regulation and governance, Slovak Ministry of Investments, Regional Development, and Information, and Chair of the Working Party on Governance of AI at the Organization for Economic Cooperation and Development.

What are the most important organizations and governing bodies when it comes to policy and governance on artificial intelligence in Europe?

Juraj Čorba

Juraj Čorba: Well, there are many. And it also creates a bit of a confusion around the globe—who are the actors in Europe? So it’s always good to clarify. First of all we have the European Union, which is a supranational organization composed of many member states, including my own Slovakia. And it was the European Union that proposed adoption of a horizontal legislation for AI in 2021. It was the initiative of the European Commission, the E.U. institution, which has a legislative initiative in the E.U. And the E.U. AI Act is now finally being adopted. It was already adopted by the European Parliament.

So this started, you said 2021. That’s before ChatGPT and the whole large language model phenomenon really took hold.

Čorba: That was the case. Well, the expert community already knew that something was being cooked in the labs. But, yes, the whole agenda of large models, including large language models, came up only later on, after 2021. So the European Union tried to reflect that. Basically, the initial proposal to regulate AI was based on a blueprint of so-called product safety, which somehow presupposes a certain intended purpose. In other words, the checks and assessments of products are based more or less on the logic of the mass production of the 20th century, on an industrial scale, right? Like when you have products that you can somehow define easily and all of them have a clearly intended purpose. Whereas with these large models, a new paradigm was arguably opened, where they have a general purpose.

So the whole proposal was then rewritten in negotiations between the Council of Ministers, which is one of the legislative bodies, and the European Parliament. And so what we have today is a combination of this old product-safety approach and some novel aspects of regulation specifically designed for what we call general-purpose artificial intelligence systems or models. So that’s the E.U.

By product safety, you mean, if AI-based software is controlling a machine, you need to have physical safety.

Čorba: Exactly. That’s one of the aspects. So that touches upon the tangible products such as vehicles, toys, medical devices, robotic arms, et cetera. So yes. But from the very beginning, the proposal contained a regulation of what the European Commission called stand-alone systems—in other words, software systems that do not necessarily command physical objects. So it was already there from the very beginning, but all of it was based on the assumption that all software has its easily identifiable intended purpose—which is not the case for general-purpose AI.

Also, large language models and generative AI in general brings in this whole other dimension, of propaganda, false information, deepfakes, and so on, which is different from traditional notions of safety in real-time software.

Čorba: Well, this is exactly the aspect that is handled by another European organization, different from the E.U., and that is the Council of Europe. It’s an international organization established after the Second World War for the protection of human rights, for protection of the rule of law, and protection of democracy. So that’s where the Europeans, but also many other states and countries, started to negotiate a first international treaty on AI. For example, the United States have participated in the negotiations, and also Canada, Japan, Australia, and many other countries. And then these particular aspects, which are related to the protection of integrity of elections, rule-of-law principles, protection of fundamental rights or human rights under international law—all these aspects have been dealt with in the context of these negotiations on the first international treaty, which is to be now adopted by the Committee of Ministers of the Council of Europe on the 16th and 17th of May. So, pretty soon. And then the first international treaty on AI will be submitted for ratifications.

So prompted largely by the activity in large language models, AI regulation and governance now is a hot topic in the United States, in Europe, and in Asia. But of the three regions, I get the sense that Europe is proceeding most aggressively on this topic of regulating and governing artificial intelligence. Do you agree that Europe is taking a more proactive stance in general than the United States and Asia?

Čorba: I’m not so sure. If you look at the Chinese approach and the way they regulate what we call generative AI, it would appear to me that they also take it very seriously. They take a different approach from the regulatory point of view. But it seems to me that, for instance, China is taking a very focused and careful approach. For the United States, I wouldn’t say that the United States is not taking a careful approach because last year you saw many of the executive orders, or even this year, some of the executive orders issued by President Biden. Of course, this was not a legislative measure, this was a presidential order. But it seems to me that the United States is also trying to address the issue very actively. The United States has also initiated the first resolution of the General Assembly at the U.N. on AI, which was passed just recently. So I wouldn’t say that the E.U. is more aggressive in comparison with Asia or North America, but maybe I would say that the E.U. is the most comprehensive. It looks horizontally across different agendas and it uses binding legislation as a tool, which is not always the case around the world. Many countries simply feel that it’s too early to legislate in a binding way, so they opt for soft measures or guidance, collaboration with private companies, et cetera. Those are the differences that I see.

Do you think you perceive a difference in focus among the three regions? Are there certain aspects that are being more aggressively pursued in the United States than in Europe or vice versa?

Čorba: Certainly the E.U. is very focused on the protection of human rights, the full catalog of human rights, but also, of course, on safety and human health. These are the core goals or values to be protected under the E.U. legislation. As for the United States and for China, I would say that the primary focus in those countries—but this is only my personal impression—is on national and economic security.

Samuel Naffziger

Samuel Naffziger is senior vice president and a corporate fellow at Advanced Micro Devices, where he is responsible for technology strategy and product architectures. Naffziger was instrumental in AMD’s embrace and development of chiplets, which are semiconductor dies that are packaged together into high-performance modules.

To what extent is large language model training starting to influence what you and your colleagues do at AMD?

Samuel Naffziger

Samuel Naffziger: Well, there are a couple levels of that. LLMs are impacting the way a lot of us live and work. And we certainly are deploying that very broadly internally for productivity enhancements, for using LLMs to provide starting points for code—simple verbal requests, such as “Give me a Python script to parse this dataset.” And you get a really nice starting point for that code. Saves a ton of time. Writing verification test benches, helping with the physical design layout optimizations. So there’s a lot of productivity aspects.

The other aspect to LLMs is, of course, we are actively involved in designing GPUs [graphics processing units] for LLM training and for LLM inference. And so that’s driving a tremendous amount of workload analysis on the requirements, hardware requirements, and hardware-software codesign, to explore.

So that brings us to your current flagship, the Instinct MI300X, which is actually billed as an AI accelerator. How did the particular demands influence that design? I don’t know when that design started, but the ChatGPT era started about two years ago or so. To what extent did you read the writing on the wall?

Naffziger: So we were just into the MI300—in 2019, we were starting the development. A long time ago. And at that time, our revenue stream from the Zen [an AMD architecture used in a family of processors] renaissance had really just started coming in. So the company was starting to get healthier, but we didn’t have a lot of extra revenue to spend on R&D at the time. So we had to be very prudent with our resources. And we had strategic engagements with the [U.S.] Department of Energy for supercomputer deployments. That was the genesis for our MI line—we were developing it for the supercomputing market. Now, there was a recognition that munching through FP64 COBOL code, or Fortran, isn’t the future, right? [laughs] This machine-learning [ML] thing is really getting some legs.

So we put some of the lower-precision math formats in, like Brain Floating Point 16 at the time, that were going to be important for inference. And the DOE knew that machine learning was going to be an important dimension of supercomputers, not just legacy code. So that’s the way, but we were focused on HPC [high-performance computing]. We had the foresight to understand that ML had real potential. Although certainly no one predicted, I think, the explosion we’ve seen today.

So that’s how it came about. And, just another piece of it: We leveraged our modular chiplet expertise to architect the 300 to support a number of variants from the same silicon components. So the variant targeted to the supercomputer market had CPUs integrated in as chiplets, directly on the silicon module. And then it had six of the GPU chiplets we call XCDs around them. So we had three CPU chiplets and six GPU chiplets. And that provided an amazingly efficient, highly integrated, CPU-plus-GPU design we call MI300A. It’s very compelling for the El Capitan supercomputer that’s being brought up as we speak.

But we also recognize that for the maximum computation for these AI workloads, the CPUs weren’t that beneficial. We wanted more GPUs. For these workloads, it’s all about the math and matrix multiplies. So we were able to just swap out those three CPU chiplets for a couple more XCD GPUs. And so we got eight XCDs in the module, and that’s what we call the MI300X. So we kind of got lucky having the right product at the right time, but there was also a lot of skill involved in that we saw the writing on the wall for where these workloads were going and we provisioned the design to support it.

Earlier you mentioned 3D chiplets. What do you feel is the next natural step in that evolution?

Naffziger: AI has created this bottomless thirst for more compute [power]. And so we are always going to be wanting to cram as many transistors as possible into a module. And the reason that’s beneficial is, these systems deliver AI performance at scale with thousands, tens of thousands, or more, compute devices. They all have to be tightly connected together, with very high bandwidths, and all of that bandwidth requires power, requires very expensive infrastructure. So if a certain level of performance is required—a certain number of petaflops, or exaflops—the strongest lever on the cost and the power consumption is the number of GPUs required to achieve a zettaflop, for instance. And if the GPU is a lot more capable, then all of that system infrastructure collapses down—if you only need half as many GPUs, everything else goes down by half. So there’s a strong economic motivation to achieve very high levels of integration and performance at the device level. And the only way to do that is with chiplets and with 3D stacking. So we’ve already embarked down that path. A lot of tough engineering problems to solve to get there, but that’s going to continue.

And so what’s going to happen? Well, obviously we can add layers, right? We can pack more in. The thermal challenges that come along with that are going to be fun engineering problems that our industry is good at solving.

by Kathy Pretz on 05. June 2024. at 18:00

To help find ways to solve transportation issues such as poorly maintained roads, traffic jams, and the high rate of accidents, researchers need access to the most current datasets on a variety of topics. But tracking down information about roadway conditions, congestion, and other statistics across multiple websites can be time-consuming. Plus, the data isn’t always accurate.

The new National Transportation Data & Analytics Solution (NTDAS), developed with the help of IEEE, makes it easier to retrieve, visualize, and analyze data in one place. NTDAS combines advanced research tools with access to high-quality transportation datasets from the U.S. Federal Highway Administration’s National Highway System and the entire Traffic Message Channel network, which distributes information on more than 1 million road segments. Anonymous data on millions of cars and trucks is generated from vehicle probes, which are vehicles equipped with GPS or global navigation satellite systems that gather traffic data on location, speed, and direction. This information helps transportation planners improve traffic flow, make transportation networks more efficient, and plan budgets.

The platform is updated monthly and contains archival data back to 2017.

“The difference between NTDAS and other competitors is that our data comes from a trusted source that means the most: the U.S. Federal Highway Administration,” says Lavanya Sayam, senior manager of data analytics alliances and programs for IEEE Global Products and Marketing. “The data has been authenticated and validated. The ability to download this massive dataset provides an unparalleled ease to data scientists and machine-learning engineers to explore and innovate.”

IEEE is diversifying its line of products beyond its traditional fields of electrical engineering, Sayam adds. “We are not just focused on electrical or computer science,” she says. “IEEE is so diverse, and this state-of-the-art platform reflects that.”

Robust analytical tools

NTDAS was built in partnership with INRIX, a transportation analytics solutions provider, and the University of Maryland’s Center for Advanced Transportation Technology Laboratory, a leader in transportation science research. INRIX provided the data, while UMD built the analytics tools. The platform leverages the National Performance Management Research Data Set, a highly granular data source from the Federal Highway Administration.

The suite of tools allows users to do tasks such as creating a personal dashboard to monitor traffic conditions on specific roads, downloading raw data for analysis, building animated maps of road conditions, and measuring the flow of traffic. There are tutorials available on the platform on how to use each tool, and templates for creating reports, documents, and pamphlets.

“The difference between National Transportation Data & Analytics Solutions and other competitors is that our data comes from a trusted source that means the most: the U.S. Federal Highway Administration.” —Lavanya Sayam

“This is the first time this type of platform is being offered by IEEE to the global academic institutional audience,” she says. “IEEE is always looking for new ways to serve the engineering community.”

A subscription-based service, NTDAS has multidisciplinary relevance, Sayam says. The use cases it includes serve researchers and educators who need a robust platform that has all the data that helps them conduct analytics in one place, she says. For university instructors, it’s an innovative way to teach the courses, and for students, it’s a unique way to apply what they’ve learned with real-world data and uses.

The platform goes beyond just those working in transportation, Sayam notes. Others who might find NTDAS useful include those who study traffic as it relates to sustainability, the environment, civil engineering, public policy, business, and logistics, she adds.

50 ways to minimize the impact of traffic

NTDAS also includes more than 50 use cases created by IEEE experts to demonstrate how the data could be analyzed. The examples identify ways to protect the environment, better serve disadvantaged communities, support alternative transportation, and improve the safety of citizens. “Data from NTDAS can be easily extrapolated to non-U.S. geographies, making it highly relevant to global researchers,” according to Sayam. This is explained in specific use cases too.

The cases cover topics such as the impact of traffic on bird populations, air-quality issues in underserved communities, and optimal areas to install electric vehicle charging stations.

Two experts covered various strategies for how to use the data to analyze the impact of transportation and infrastructure on the environment in this on-demand webinar held in May.

Thomas Brennan, a professor of civil engineering at the College of New Jersey, discussed how using NTDAS data could aid in better planning of evacuation routes during wildfires, such as determining the location of first responders and traffic congestion in the area, including seasonal traffic. This and other data could lead to evacuating residents faster, new evacuation road signage, and better communication warning systems, he said.

“Traffic systems are super complex and very difficult to understand and model,” said presenter Jane MacFarlane, director of the Smart Cities and Sustainable Mobility Center at the University of California’s Institute of Transportation Studies, in Berkeley. “Now that we have datasets like these, that’s giving us a huge leg up in trying to use them for predictive modeling and also helping us with simulating things so that we can gain a better understanding.”

Watch this short demonstration about the National Transportation Data & Analytics Solutions platform.

“Transportation is a basic fabric of society,” Sayam says. “Understanding its impact is an imperative for better living. True to IEEE’s mission of advancing technology for humanity, NTDAS, with its interdisciplinary relevance, helps us understand the impact of transportation across several dimensions.”

by Tekla S. Perry on 04. June 2024. at 13:06

Elemind, a 5-year-old startup based in Cambridge, Mass., today unveiled a US $349 wearable for neuromodulation, the company’s first product. According to cofounder and CEO Meredith Perry, the technology tracks the oscillation of brain waves using electroencephalography (EEG) sensors that detect the electrical activity of the brain and then influence those oscillations using bursts of sound delivered via bone conduction.

Elemind’s first application for this wearable aims to suppress alpha waves to help induce sleep. There are other wearables on the market that monitor brain waves and, through biofeedback, encourage users to actively modify their alpha patterns. Elemind’s headband appears to be the first device to use sound to directly influence the brain waves of a passive user.

In a clinical trial, says Perry [no relation to author], 76 percent of subjects fell asleep more quickly. Those who did see a difference averaged 48 percent less time to progress from awake to asleep. The results were similar to those of comparable trials of pharmaceutical sleep aids, Perry indicated.

“For me,” Perry said, “it cuts through my rumination, quiets my thinking. It’s like noise cancellation for the brain.”

I briefly tested Elemind’s headband in May. I found it comfortable, with a thick cushioned band that sits across the forehead connected to a stretchy elastic loop to keep it in place. In the band are multiple EEG electrodes, a processor, a three-axis accelerometer, a rechargeable lithium-polymer battery, and custom electronics that gather the brain’s electrical signals, estimate their phase, and generate pink noise through a bone-conduction speaker. The whole thing weighs about 60 grams—about as much as a small kiwi fruit.

My test conditions were far from optimal for sleep: early afternoon, a fairly bright conference room, a beanbag chair as bed, and a vent blowing. And my test lasted just 4 minutes. I can say that I didn’t find the little bursts of pink noise (white noise without the higher frequencies) unpleasant. And since I often wear an eye mask, feeling fabric on my face wasn’t disturbing. It wasn’t the time or place to try for sound sleep, but I—and the others in the room—noted that after 2 minutes I was yawning like crazy.

How Elemind tweaks brain waves

What was going on in my brain? Briefly, different brain states are associated with different frequencies of waves. Someone who is relaxed with eyes closed but not asleep produces alpha waves at around 10 hertz. As they drift off to sleep, the alpha waves are supplanted by theta waves, at around 5 Hz. Eventually, the delta waves of deep sleep show up at around 1 Hz.

Ryan Neely, Elemind’s vice president of science and research, explains: “As soon as you put the headband on,” he says, “the EEG system starts running. It uses straightforward signal processing with bandpass filtering to isolate the activity in the 8- to 12-Hz frequency range—the alpha band.”

“Then,” Neely continues, “our algorithm looks at the filtered signal to identify the phase of each oscillation and determines when to generate bursts of pink noise.”

To help a user fall asleep more quickly [top], bursts of pink noise are timed to generate a brain response that is out of phase with alpha waves and so suppresses them. To enhance deep sleep [bottom], the pink noise is timed to generate a brain response that is in phase with delta waves.Source: Elemind

These auditory stimuli, he explains, create ripples in the waves coming from the brain. Elemind’s system tries to align these ripples with a particular phase in the wave. Because there is a gap between the stimulus and the evoked response, Elemind tested its system on 21 people and calculated the average delay, taking that into account when determining when to trigger a sound.

To induce sleep, Elemind’s headband targets the trough in the alpha wave, the point at which the brain is most excitable, Neely says.

“You can think of the alpha rhythm as a gate for communication between different areas of the brain,” he says. “By interfering with that communication, that coordination between different brain areas, you can disrupt patterns, like the ruminations that keep you awake.”

With these alpha waves suppressed, Neely says, the slower oscillations, like the theta waves of light sleep, take over.

Elemind doesn’t plan to stop there. The company plans to add an algorithm that addresses delta waves, the low-frequency 0.5- to 2-Hz waves characteristic of deep sleep. Here, Elemind’s technology will attempt to amplify this pattern with the intent of improving sleep quality.

Is this safe? Yes, Neely says, because auditory stimulation is self-limiting. “Your brain waves have a natural space they can occupy,” he explains, “and this stimulation just moved it within that natural space, unlike deep-brain stimulation, which can move the brain activity outside natural parameters.”

Going beyond sleep to sedation, memory, and mental health

Applications may eventually go beyond inducing and enhancing sleep. Researchers at the University of Washington and McGill University have completed a clinical study to determine if Elemind’s technology can be used to increase the pain threshold of subjects undergoing sedation. The results are being prepared for peer review.

Elemind is also working with a team involving researchers at McGill and the Leuven Brain Institute to determine if the technology can enhance memory consolidation in deep sleep and perhaps have some usefulness for people with mild cognitive impairment and other memory disorders.

Neely would love to see more applications investigated in the future.

“Inverse alpha stimulation [enhancing instead of suppressing the signal] could increase arousal,” he says. “That’s something I’d love to look into. And looking into mental-health treatment would be interesting, because phase coupling between the different brain regions appears to be an important factor in depression and anxiety disorders.”

Perry, who previously founded the wireless power startup UBeam, cofounded Elemind with four university professors with expertise in neuroscience, optogenetics, biomedical engineering, and artificial intelligence. The company has $12 million in funding to date and currently has 13 employees.

Preorders at $349 start today for beta units, and Elemind expects to start general sales later this year. The company will offer customers an optional membership at $7 to $13 monthly that will allow cloud storage of sleep data and access to new apps as they are released.

by Eliza Strickland on 09. February 2022. at 15:31

Andrew Ng has serious street cred in artificial intelligence. He pioneered the use of graphics processing units (GPUs) to train deep learning models in the late 2000s with his students at Stanford University, cofounded Google Brain in 2011, and then served for three years as chief scientist for Baidu, where he helped build the Chinese tech giant’s AI group. So when he says he has identified the next big shift in artificial intelligence, people listen. And that’s what he told IEEE Spectrum in an exclusive Q&A.

---

Ng’s current efforts are focused on his company Landing AI, which built a platform called LandingLens to help manufacturers improve visual inspection with computer vision. He has also become something of an evangelist for what he calls the data-centric AI movement, which he says can yield “small data” solutions to big issues in AI, including model efficiency, accuracy, and bias.

Andrew Ng on...

• What’s next for really big models
• The career advice he didn’t listen to
• Defining the data-centric AI movement
• Synthetic data
• Why Landing AI asks its customers to do the work

The great advances in deep learning over the past decade or so have been powered by ever-bigger models crunching ever-bigger amounts of data. Some people argue that that’s an unsustainable trajectory. Do you agree that it can’t go on that way?

Andrew Ng: This is a big question. We’ve seen foundation models in NLP [natural language processing]. I’m excited about NLP models getting even bigger, and also about the potential of building foundation models in computer vision. I think there’s lots of signal to still be exploited in video: We have not been able to build foundation models yet for video because of compute bandwidth and the cost of processing video, as opposed to tokenized text. So I think that this engine of scaling up deep learning algorithms, which has been running for something like 15 years now, still has steam in it. Having said that, it only applies to certain problems, and there’s a set of other problems that need small data solutions.

When you say you want a foundation model for computer vision, what do you mean by that?

Ng: This is a term coined by Percy Liang and some of my friends at Stanford to refer to very large models, trained on very large data sets, that can be tuned for specific applications. For example, GPT-3 is an example of a foundation model [for NLP]. Foundation models offer a lot of promise as a new paradigm in developing machine learning applications, but also challenges in terms of making sure that they’re reasonably fair and free from bias, especially if many of us will be building on top of them.

What needs to happen for someone to build a foundation model for video?

Ng: I think there is a scalability problem. The compute power needed to process the large volume of images for video is significant, and I think that’s why foundation models have arisen first in NLP. Many researchers are working on this, and I think we’re seeing early signs of such models being developed in computer vision. But I’m confident that if a semiconductor maker gave us 10 times more processor power, we could easily find 10 times more video to build such models for vision.

Having said that, a lot of what’s happened over the past decade is that deep learning has happened in consumer-facing companies that have large user bases, sometimes billions of users, and therefore very large data sets. While that paradigm of machine learning has driven a lot of economic value in consumer software, I find that that recipe of scale doesn’t work for other industries.

Back to top

It’s funny to hear you say that, because your early work was at a consumer-facing company with millions of users.

Ng: Over a decade ago, when I proposed starting the Google Brain project to use Google’s compute infrastructure to build very large neural networks, it was a controversial step. One very senior person pulled me aside and warned me that starting Google Brain would be bad for my career. I think he felt that the action couldn’t just be in scaling up, and that I should instead focus on architecture innovation.

“In many industries where giant data sets simply don’t exist, I think the focus has to shift from big data to good data. Having 50 thoughtfully engineered examples can be sufficient to explain to the neural network what you want it to learn.”
—Andrew Ng, CEO & Founder, Landing AI

I remember when my students and I published the first NeurIPS workshop paper advocating using CUDA, a platform for processing on GPUs, for deep learning—a different senior person in AI sat me down and said, “CUDA is really complicated to program. As a programming paradigm, this seems like too much work.” I did manage to convince him; the other person I did not convince.

I expect they’re both convinced now.

Ng: I think so, yes.

Over the past year as I’ve been speaking to people about the data-centric AI movement, I’ve been getting flashbacks to when I was speaking to people about deep learning and scalability 10 or 15 years ago. In the past year, I’ve been getting the same mix of “there’s nothing new here” and “this seems like the wrong direction.”

Back to top

How do you define data-centric AI, and why do you consider it a movement?

Ng: Data-centric AI is the discipline of systematically engineering the data needed to successfully build an AI system. For an AI system, you have to implement some algorithm, say a neural network, in code and then train it on your data set. The dominant paradigm over the last decade was to download the data set while you focus on improving the code. Thanks to that paradigm, over the last decade deep learning networks have improved significantly, to the point where for a lot of applications the code—the neural network architecture—is basically a solved problem. So for many practical applications, it’s now more productive to hold the neural network architecture fixed, and instead find ways to improve the data.

When I started speaking about this, there were many practitioners who, completely appropriately, raised their hands and said, “Yes, we’ve been doing this for 20 years.” This is the time to take the things that some individuals have been doing intuitively and make it a systematic engineering discipline.

The data-centric AI movement is much bigger than one company or group of researchers. My collaborators and I organized a data-centric AI workshop at NeurIPS, and I was really delighted at the number of authors and presenters that showed up.

You often talk about companies or institutions that have only a small amount of data to work with. How can data-centric AI help them?

Ng: You hear a lot about vision systems built with millions of images—I once built a face recognition system using 350 million images. Architectures built for hundreds of millions of images don’t work with only 50 images. But it turns out, if you have 50 really good examples, you can build something valuable, like a defect-inspection system. In many industries where giant data sets simply don’t exist, I think the focus has to shift from big data to good data. Having 50 thoughtfully engineered examples can be sufficient to explain to the neural network what you want it to learn.

When you talk about training a model with just 50 images, does that really mean you’re taking an existing model that was trained on a very large data set and fine-tuning it? Or do you mean a brand new model that’s designed to learn only from that small data set?

Ng: Let me describe what Landing AI does. When doing visual inspection for manufacturers, we often use our own flavor of RetinaNet. It is a pretrained model. Having said that, the pretraining is a small piece of the puzzle. What’s a bigger piece of the puzzle is providing tools that enable the manufacturer to pick the right set of images [to use for fine-tuning] and label them in a consistent way. There’s a very practical problem we’ve seen spanning vision, NLP, and speech, where even human annotators don’t agree on the appropriate label. For big data applications, the common response has been: If the data is noisy, let’s just get a lot of data and the algorithm will average over it. But if you can develop tools that flag where the data’s inconsistent and give you a very targeted way to improve the consistency of the data, that turns out to be a more efficient way to get a high-performing system.

“Collecting more data often helps, but if you try to collect more data for everything, that can be a very expensive activity.”
—Andrew Ng

For example, if you have 10,000 images where 30 images are of one class, and those 30 images are labeled inconsistently, one of the things we do is build tools to draw your attention to the subset of data that’s inconsistent. So you can very quickly relabel those images to be more consistent, and this leads to improvement in performance.

Could this focus on high-quality data help with bias in data sets? If you’re able to curate the data more before training?

Ng: Very much so. Many researchers have pointed out that biased data is one factor among many leading to biased systems. There have been many thoughtful efforts to engineer the data. At the NeurIPS workshop, Olga Russakovsky gave a really nice talk on this. At the main NeurIPS conference, I also really enjoyed Mary Gray’s presentation, which touched on how data-centric AI is one piece of the solution, but not the entire solution. New tools like Datasheets for Datasets also seem like an important piece of the puzzle.

One of the powerful tools that data-centric AI gives us is the ability to engineer a subset of the data. Imagine training a machine-learning system and finding that its performance is okay for most of the data set, but its performance is biased for just a subset of the data. If you try to change the whole neural network architecture to improve the performance on just that subset, it’s quite difficult. But if you can engineer a subset of the data you can address the problem in a much more targeted way.

When you talk about engineering the data, what do you mean exactly?

Ng: In AI, data cleaning is important, but the way the data has been cleaned has often been in very manual ways. In computer vision, someone may visualize images through a Jupyter notebook and maybe spot the problem, and maybe fix it. But I’m excited about tools that allow you to have a very large data set, tools that draw your attention quickly and efficiently to the subset of data where, say, the labels are noisy. Or to quickly bring your attention to the one class among 100 classes where it would benefit you to collect more data. Collecting more data often helps, but if you try to collect more data for everything, that can be a very expensive activity.

For example, I once figured out that a speech-recognition system was performing poorly when there was car noise in the background. Knowing that allowed me to collect more data with car noise in the background, rather than trying to collect more data for everything, which would have been expensive and slow.

Back to top

What about using synthetic data, is that often a good solution?

Ng: I think synthetic data is an important tool in the tool chest of data-centric AI. At the NeurIPS workshop, Anima Anandkumar gave a great talk that touched on synthetic data. I think there are important uses of synthetic data that go beyond just being a preprocessing step for increasing the data set for a learning algorithm. I’d love to see more tools to let developers use synthetic data generation as part of the closed loop of iterative machine learning development.

Do you mean that synthetic data would allow you to try the model on more data sets?

Ng: Not really. Here’s an example. Let’s say you’re trying to detect defects in a smartphone casing. There are many different types of defects on smartphones. It could be a scratch, a dent, pit marks, discoloration of the material, other types of blemishes. If you train the model and then find through error analysis that it’s doing well overall but it’s performing poorly on pit marks, then synthetic data generation allows you to address the problem in a more targeted way. You could generate more data just for the pit-mark category.

“In the consumer software Internet, we could train a handful of machine-learning models to serve a billion users. In manufacturing, you might have 10,000 manufacturers building 10,000 custom AI models.”
—Andrew Ng

Synthetic data generation is a very powerful tool, but there are many simpler tools that I will often try first. Such as data augmentation, improving labeling consistency, or just asking a factory to collect more data.

Back to top

To make these issues more concrete, can you walk me through an example? When a company approaches Landing AI and says it has a problem with visual inspection, how do you onboard them and work toward deployment?

Ng: When a customer approaches us we usually have a conversation about their inspection problem and look at a few images to verify that the problem is feasible with computer vision. Assuming it is, we ask them to upload the data to the LandingLens platform. We often advise them on the methodology of data-centric AI and help them label the data.

One of the foci of Landing AI is to empower manufacturing companies to do the machine learning work themselves. A lot of our work is making sure the software is fast and easy to use. Through the iterative process of machine learning development, we advise customers on things like how to train models on the platform, when and how to improve the labeling of data so the performance of the model improves. Our training and software supports them all the way through deploying the trained model to an edge device in the factory.

How do you deal with changing needs? If products change or lighting conditions change in the factory, can the model keep up?

Ng: It varies by manufacturer. There is data drift in many contexts. But there are some manufacturers that have been running the same manufacturing line for 20 years now with few changes, so they don’t expect changes in the next five years. Those stable environments make things easier. For other manufacturers, we provide tools to flag when there’s a significant data-drift issue. I find it really important to empower manufacturing customers to correct data, retrain, and update the model. Because if something changes and it’s 3 a.m. in the United States, I want them to be able to adapt their learning algorithm right away to maintain operations.

In the consumer software Internet, we could train a handful of machine-learning models to serve a billion users. In manufacturing, you might have 10,000 manufacturers building 10,000 custom AI models. The challenge is, how do you do that without Landing AI having to hire 10,000 machine learning specialists?

So you’re saying that to make it scale, you have to empower customers to do a lot of the training and other work.

Ng: Yes, exactly! This is an industry-wide problem in AI, not just in manufacturing. Look at health care. Every hospital has its own slightly different format for electronic health records. How can every hospital train its own custom AI model? Expecting every hospital’s IT personnel to invent new neural-network architectures is unrealistic. The only way out of this dilemma is to build tools that empower the customers to build their own models by giving them tools to engineer the data and express their domain knowledge. That’s what Landing AI is executing in computer vision, and the field of AI needs other teams to execute this in other domains.

Is there anything else you think it’s important for people to understand about the work you’re doing or the data-centric AI movement?

Ng: In the last decade, the biggest shift in AI was a shift to deep learning. I think it’s quite possible that in this decade the biggest shift will be to data-centric AI. With the maturity of today’s neural network architectures, I think for a lot of the practical applications the bottleneck will be whether we can efficiently get the data we need to develop systems that work well. The data-centric AI movement has tremendous energy and momentum across the whole community. I hope more researchers and developers will jump in and work on it.

Back to top

This article appears in the April 2022 print issue as “Andrew Ng, AI Minimalist.”

by Rina Diane Caballar on 08. February 2022. at 14:00

The end of Moore’s Law is looming. Engineers and designers can do only so much to miniaturize transistors and pack as many of them as possible into chips. So they’re turning to other approaches to chip design, incorporating technologies like AI into the process.

Samsung, for instance, is adding AI to its memory chips to enable processing in memory, thereby saving energy and speeding up machine learning. Speaking of speed, Google’s TPU V4 AI chip has doubled its processing power compared with that of its previous version.

But AI holds still more promise and potential for the semiconductor industry. To better understand how AI is set to revolutionize chip design, we spoke with Heather Gorr, senior product manager for MathWorks’ MATLAB platform.

How is AI currently being used to design the next generation of chips?

Heather Gorr: AI is such an important technology because it’s involved in most parts of the cycle, including the design and manufacturing process. There’s a lot of important applications here, even in the general process engineering where we want to optimize things. I think defect detection is a big one at all phases of the process, especially in manufacturing. But even thinking ahead in the design process, [AI now plays a significant role] when you’re designing the light and the sensors and all the different components. There’s a lot of anomaly detection and fault mitigation that you really want to consider.

Heather GorrMathWorks

Then, thinking about the logistical modeling that you see in any industry, there is always planned downtime that you want to mitigate; but you also end up having unplanned downtime. So, looking back at that historical data of when you’ve had those moments where maybe it took a bit longer than expected to manufacture something, you can take a look at all of that data and use AI to try to identify the proximate cause or to see something that might jump out even in the processing and design phases. We think of AI oftentimes as a predictive tool, or as a robot doing something, but a lot of times you get a lot of insight from the data through AI.

What are the benefits of using AI for chip design?

Gorr: Historically, we’ve seen a lot of physics-based modeling, which is a very intensive process. We want to do a reduced order model, where instead of solving such a computationally expensive and extensive model, we can do something a little cheaper. You could create a surrogate model, so to speak, of that physics-based model, use the data, and then do your parameter sweeps, your optimizations, your Monte Carlo simulations using the surrogate model. That takes a lot less time computationally than solving the physics-based equations directly. So, we’re seeing that benefit in many ways, including the efficiency and economy that are the results of iterating quickly on the experiments and the simulations that will really help in the design.

So it’s like having a digital twin in a sense?

Gorr: Exactly. That’s pretty much what people are doing, where you have the physical system model and the experimental data. Then, in conjunction, you have this other model that you could tweak and tune and try different parameters and experiments that let sweep through all of those different situations and come up with a better design in the end.

So, it’s going to be more efficient and, as you said, cheaper?

Gorr: Yeah, definitely. Especially in the experimentation and design phases, where you’re trying different things. That’s obviously going to yield dramatic cost savings if you’re actually manufacturing and producing [the chips]. You want to simulate, test, experiment as much as possible without making something using the actual process engineering.

We’ve talked about the benefits. How about the drawbacks?

Gorr: The [AI-based experimental models] tend to not be as accurate as physics-based models. Of course, that’s why you do many simulations and parameter sweeps. But that’s also the benefit of having that digital twin, where you can keep that in mind—it’s not going to be as accurate as that precise model that we’ve developed over the years.

Both chip design and manufacturing are system intensive; you have to consider every little part. And that can be really challenging. It’s a case where you might have models to predict something and different parts of it, but you still need to bring it all together.

One of the other things to think about too is that you need the data to build the models. You have to incorporate data from all sorts of different sensors and different sorts of teams, and so that heightens the challenge.

How can engineers use AI to better prepare and extract insights from hardware or sensor data?

Gorr: We always think about using AI to predict something or do some robot task, but you can use AI to come up with patterns and pick out things you might not have noticed before on your own. People will use AI when they have high-frequency data coming from many different sensors, and a lot of times it’s useful to explore the frequency domain and things like data synchronization or resampling. Those can be really challenging if you’re not sure where to start.

One of the things I would say is, use the tools that are available. There’s a vast community of people working on these things, and you can find lots of examples [of applications and techniques] on GitHub or MATLAB Central, where people have shared nice examples, even little apps they’ve created. I think many of us are buried in data and just not sure what to do with it, so definitely take advantage of what’s already out there in the community. You can explore and see what makes sense to you, and bring in that balance of domain knowledge and the insight you get from the tools and AI.

What should engineers and designers consider when using AI for chip design?

Gorr: Think through what problems you’re trying to solve or what insights you might hope to find, and try to be clear about that. Consider all of the different components, and document and test each of those different parts. Consider all of the people involved, and explain and hand off in a way that is sensible for the whole team.

How do you think AI will affect chip designers’ jobs?

Gorr: It’s going to free up a lot of human capital for more advanced tasks. We can use AI to reduce waste, to optimize the materials, to optimize the design, but then you still have that human involved whenever it comes to decision-making. I think it’s a great example of people and technology working hand in hand. It’s also an industry where all people involved—even on the manufacturing floor—need to have some level of understanding of what’s happening, so this is a great industry for advancing AI because of how we test things and how we think about them before we put them on the chip.

How do you envision the future of AI and chip design?

Gorr: It’s very much dependent on that human element—involving people in the process and having that interpretable model. We can do many things with the mathematical minutiae of modeling, but it comes down to how people are using it, how everybody in the process is understanding and applying it. Communication and involvement of people of all skill levels in the process are going to be really important. We’re going to see less of those superprecise predictions and more transparency of information, sharing, and that digital twin—not only using AI but also using our human knowledge and all of the work that many people have done over the years.

by Dexter Johnson on 07. February 2022. at 16:12

Quantum computing is a devilishly complex technology, with many technical hurdles impacting its development. Of these challenges two critical issues stand out: miniaturization and qubit quality.

IBM has adopted the superconducting qubit road map of reaching a 1,121-qubit processor by 2023, leading to the expectation that 1,000 qubits with today’s qubit form factor is feasible. However, current approaches will require very large chips (50 millimeters on a side, or larger) at the scale of small wafers, or the use of chiplets on multichip modules. While this approach will work, the aim is to attain a better path toward scalability.

Now researchers at MIT have been able to both reduce the size of the qubits and done so in a way that reduces the interference that occurs between neighboring qubits. The MIT researchers have increased the number of superconducting qubits that can be added onto a device by a factor of 100.

“We are addressing both qubit miniaturization and quality,” said William Oliver, the director for the Center for Quantum Engineering at MIT. “Unlike conventional transistor scaling, where only the number really matters, for qubits, large numbers are not sufficient, they must also be high-performance. Sacrificing performance for qubit number is not a useful trade in quantum computing. They must go hand in hand.”

The key to this big increase in qubit density and reduction of interference comes down to the use of two-dimensional materials, in particular the 2D insulator hexagonal boron nitride (hBN). The MIT researchers demonstrated that a few atomic monolayers of hBN can be stacked to form the insulator in the capacitors of a superconducting qubit.

Just like other capacitors, the capacitors in these superconducting circuits take the form of a sandwich in which an insulator material is sandwiched between two metal plates. The big difference for these capacitors is that the superconducting circuits can operate only at extremely low temperatures—less than 0.02 degrees above absolute zero (-273.15 °C).

Superconducting qubits are measured at temperatures as low as 20 millikelvin in a dilution refrigerator.Nathan Fiske/MIT

In that environment, insulating materials that are available for the job, such as PE-CVD silicon oxide or silicon nitride, have quite a few defects that are too lossy for quantum computing applications. To get around these material shortcomings, most superconducting circuits use what are called coplanar capacitors. In these capacitors, the plates are positioned laterally to one another, rather than on top of one another.

As a result, the intrinsic silicon substrate below the plates and to a smaller degree the vacuum above the plates serve as the capacitor dielectric. Intrinsic silicon is chemically pure and therefore has few defects, and the large size dilutes the electric field at the plate interfaces, all of which leads to a low-loss capacitor. The lateral size of each plate in this open-face design ends up being quite large (typically 100 by 100 micrometers) in order to achieve the required capacitance.

In an effort to move away from the large lateral configuration, the MIT researchers embarked on a search for an insulator that has very few defects and is compatible with superconducting capacitor plates.

“We chose to study hBN because it is the most widely used insulator in 2D material research due to its cleanliness and chemical inertness,” said colead author Joel Wang, a research scientist in the Engineering Quantum Systems group of the MIT Research Laboratory for Electronics.

On either side of the hBN, the MIT researchers used the 2D superconducting material, niobium diselenide. One of the trickiest aspects of fabricating the capacitors was working with the niobium diselenide, which oxidizes in seconds when exposed to air, according to Wang. This necessitates that the assembly of the capacitor occur in a glove box filled with argon gas.

While this would seemingly complicate the scaling up of the production of these capacitors, Wang doesn’t regard this as a limiting factor.

“What determines the quality factor of the capacitor are the two interfaces between the two materials,” said Wang. “Once the sandwich is made, the two interfaces are “sealed” and we don’t see any noticeable degradation over time when exposed to the atmosphere.”

This lack of degradation is because around 90 percent of the electric field is contained within the sandwich structure, so the oxidation of the outer surface of the niobium diselenide does not play a significant role anymore. This ultimately makes the capacitor footprint much smaller, and it accounts for the reduction in cross talk between the neighboring qubits.

“The main challenge for scaling up the fabrication will be the wafer-scale growth of hBN and 2D superconductors like [niobium diselenide], and how one can do wafer-scale stacking of these films,” added Wang.

Wang believes that this research has shown 2D hBN to be a good insulator candidate for superconducting qubits. He says that the groundwork the MIT team has done will serve as a road map for using other hybrid 2D materials to build superconducting circuits.