IEEE Spectrum IEEE Spectrum
-
Video Friday: Disney Robot Dance
by Evan Ackerman on 23. August 2024. at 17:30
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.
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!
I think it’s time for us all to admit that some of the most interesting bipedal and humanoid research is being done by Disney.
[ Research Paper from ETH Zurich and Disney Research]
Over the past few months, Unitree G1 robot has been upgraded into a mass production version, with stronger performance, ultimate appearance, and being more in line with mass production requirements.
[ Unitree ]
This robot is from Kinisi Robotics, which was founded by Brennand Pierce, who also founded Bear Robotics. You can’t really tell from this video, but check out the website because the reach this robot has is bonkers.
Kinisi Robotics is on a mission to democratize access to advanced robotics with our latest innovation—a low-cost, dual-arm robot designed for warehouses, factories, and supermarkets. What sets our robot apart is its integration of LLM technology, enabling it to learn from demonstrations and perform complex tasks with minimal setup. Leveraging Brennand’s extensive experience in scaling robotic solutions, we’re able to produce this robot for under $20k, making it a game-changer in the industry.
[ Kinisi Robotics ]
Thanks Bren!
Finally, something that Atlas does that I am also physically capable of doing. In theory.
Okay, never mind. I don’t have those hips.
[ Boston Dynamics ]
Researchers in the Department of Mechanical Engineering at Carnegie Mellon University have created the first legged robot of its size to run, turn, push loads, and climb miniature stairs.
They say it can “run,” but I’m skeptical that there’s a flight phase unless someone sneezes nearby.
[ Carnegie Mellon University ]
The lights are cool and all, but it’s the pulsing soft skin that’s squigging me out.
[ Paper, Robotics Reports Vol.2 ]
Roofing is a difficult and dangerous enough job that it would be great if robots could take it over. It’ll be a challenge though.
[ Renovate Robotics ] via [ TechCrunch ]
Kento Kawaharazuka from JSK Robotics Laboratory at the University of Tokyo wrote in to share this paper, just accepted at RA-L, which (among other things) shows a robot using its flexible hands to identify objects through random finger motion.
[ Paper accepted by IEEE Robotics and Automation Letters ]
Thanks Kento!
It’s one thing to make robots that are reliable, and it’s another to make robots that are reliable and repairable by the end user. I don’t think iRobot gets enough credit for this.
[ iRobot ]
I like competitions where they say, “just relax and forget about the competition and show us what you can do.”
[ MBZIRC Maritime Grand Challenge ]
I kid you not, this used to be my job.
[ RoboHike ]
-
Early-Career Telecom Engineer Is a Model of IEEE Engagement
by Willie D. Jones on 22. August 2024. at 18:00
There were early signs that Muhammad Hamza Ihtisham was born to excel in engineering or computer science, but family tradition initially steered him toward a career in medicine. Ihtisham’s mother and other family members were medical professionals. His father is a businessman.
Even though his father had dreamed of having a son in the engineering field, it was assumed that Ihtisham would become a doctor. And he almost did. But when he didn’t pass the medical qualifying exams in high school, he saw it as a sign to switch professions.
Muhammad Hamza Ihtisham
Employer:
Jazz in Lahore, Pakistan
Title:
Network experience specialist in radio
Member grade:
IEEE member
Alma mater:
University of the Punjab
“I asked my parents and my principal for permission to switch my focus from medicine to computer science,” he says. Although the change took effect only three months before he had to take exams, he scored high enough to place third among the computer science students at his school. He has never looked back.
Ihtisham is now a network experience specialist in radio at the largest telecommunications provider in Pakistan: Lahore-based Jazz. Ihtisham monitors, supervises, and troubleshoots nationwide wireless networks and is a team player who implements smart systems and AI-based solutions to optimize network performance.
“We are working with 2G, 3G, and 4G network supervision,” he says, “and we’re also evolving for 5G and optical fiber networks.”
Muhammad Hamza Ihtisham chats with 2018 IEEE President Jim Jefferies.Muhammad Hamza Ihtisham
Early evidence of STEM affinity
Ihtisham didn’t need much encouragement to become an engineer, he says, adding that he always has wanted to work with technology.
When he was young, his father bought him a computer with a Pentium II CPU “at a time when there were very few computers in my town,” he says.
Ihtisham’s curiosity led him to dismantle and explore its components, fostering a deeper interest in technology.
“I destroyed many motherboards and processors when I pulled them out of the computer to see how they worked,” he says. It was part of his innate tendency to get to the bottom of how things worked. “I still have that spark in me, that inner child who wants to open things up and investigate how they work.”
He earned a bachelor’s degree in electrical engineering with a specialization in telecommunications from the University of the Punjab in 2018. He returned to pursue a master’s degree in industrial engineering and management—which he received in 2022.
Thanks to his IEEE connections, Ihtisham secured his position at Jazz even before getting his bachelor’s degree. As the university’s IEEE student branch chair, Ihtisham invited an engineer from Jazz to speak to students on campus. When Ihtisham later showed up at the company for a job interview, that same engineer was the department head and immediately recognized him.
Since his college days, Ihtisham has poured time and energy into giving back to the profession through participation in IEEE. He founded his school’s student chapter and today serves as chair of the IEEE Lahore Section’s Young Professionals group. He is also the deputy lead of the global technical and operation committee of the IEEE Young Professionals mentoring program, which connects experts with mentees to help them learn and further their career.
Active IEEE student leader
Ihtisham entered college thinking he would become a computer scientist, but before long he became convinced that his true passion lay in engineering. Noticing a gap in student activities within the school’s EE department, he joined IEEE in his third semester.
Although the university was more than 150 years old, electrical engineering was a relatively new course of study there.
“My graduating class had only the 10th cohort of graduates to earn that degree from the university,” he says.
As founder of the school’s IEEE student branch, Ihtisham set about adding activities and opportunities for would-be engineers that he felt were missing. He was the branch’s first chair, organizing activities, boosting membership, and overseeing initiatives that impacted his university and the wider IEEE Lahore Section. He was then appointed a student representative for the section.
“That was a turning point for me,” he says.
He originally started volunteering with IEEE for a pragmatic purpose that served the entire engineering student body, he says, but as he settled into his new leadership roles, volunteering became a source of personal fulfillment and development.
“When I started my IEEE journey, I was not prepared. But I worked on my leadership, my behavior, and improving my soft skills. So, you could say my involvement with IEEE has transformed my personality and served as leadership training.”
For his efforts, he has been recognized with several awards including the IEEE Lahore Section’s 2018 Outstanding Volunteer for organizing student activities and conferences.
“When I started my IEEE journey, I was not well groomed,” Ihistham says. “But I worked on my leadership, my behavior, and improving my soft skills. So, you could say my involvement with IEEE has transformed my personality and served as leadership training.”
The communications and negotiating skills he picked up by networking with IEEE members across the globe have benefited him at Jazz, he says.
His dedication to IEEE didn’t end with his student years. Today his roles involve mentoring, networking, and leading initiatives to foster growth and collaboration in the engineering community.
Now his leadership skills help him manage and motivate other volunteers and mentor engineering students. He received the 2021 IEEE MGA Young Professionals Achievement Award for organizing YP activities and the 2021 IEEE IAS Young Member Service Award for virtually engaging IEEE Industry Applications Society members during the COVID-19 pandemic.
Advice for aspiring engineers
To students considering a career in electrical engineering, Ihtisham emphasizes the importance of finding the right mentors and embracing open-source collaboration. He advises discussing ideas with experts to gain valuable insights and foster innovative thinking.
His success story underscores the value of mentorship, continuous learning, and community engagement. While he was in graduate school working toward his master’s degree, he began doing research to develop an effective and reliable brain-computer interface. He talked with the medical professionals in his family for information about how the brain works but then found himself at an impasse because there were not enough datasets in Pakistan for training his machine-learning software.
He reached out to the IEEE community and found a mentor for the project at the University of New South Wales in Sydney. Their collaboration was fruitful enough that Ihtisham was invited to present a TEDx talk on what he had learned about addiction and neurofeedback.
Based on that project, he took home third prize in the IEEE IAS Chapters and Membership Department Zucker Undergraduate Student Design Contest in 2019.
Ihtisham’s journey with IEEE exemplifies the impact of dedication, mentorship, and continued learning on building an interesting and successful engineering career.
“My success is having an impact on my younger cousins,” he says. “If they want to pursue a career in engineering or another STEM field, they have someone in the family who can guide them.”
-
Top Programming Languages Methodology 2024
by Stephen Cass on 22. August 2024. at 15:04
In our goal of trying to estimate a programming language’s popularity, we realized that no one can look over the shoulder of every person writing code, whether that be a child writing a Java script for a personal Minecraft server, a mobile app developer hoping to hit it big, or an aerospace engineer writing mission-critical code for a voyage to Mars. Our Top Programming Languages interactive tries to tackle the problem of estimating a language’s popularity by looking for proxy signals.
We do this by constructing measures of popularity from a variety of data sources that we believe are good proxies for active interest for each programming language. In total, we identify 63 programming languages. We then weight each data source to create an overall index of popularity, excluding some of the lowest scorers. Below, we describe the sources of data we use to get the measures, and the weighting scheme we use to produce the overall indices.
By popularity, we mean we are trying to rank languages that are in active use. We look at three different aspects of popularity: languages in active use among typical IEEE members and working software engineers (the “Spectrum” ranking), languages that are in demand by employers (the “Jobs” ranking), and languages that are in the zeitgeist (the “Trending” ranking).
We gauged the popularity of languages using the following sources for a total of eight metrics (see below). We gathered the information for all metrics in July—August 2024. The data were gathered manually to avoid results being biased due to API changes or terminations and because many of the programming language’s names (C++, Scheme) collided with common terms found in research papers and job ads or were difficult for a search engine to parse. When a large number of search results made it impractical to resolve ambiguities by examining all of the results individually, we used a sample of each data source, and determined the relevant sample size based on estimating the true mean with 95 percent confidence. Not all data sources contain information for each programming language and we interpret this information as the programming language having “no hits” (that is, not being popular).
The results from each metric are normalized to produce a relative popularity score between 0 and 1. Then the individual metrics are multiplied by a weight factor, combined, and the result renormalized to produce an aggregate popularity score.
In aggregating metrics, we hope to compensate for statistical quirks that might distort a language’s popularity score in any particular source of data. Varying the weight factors allows us to create the different results for the Spectrum, Jobs, and Trending rankings. We fully acknowledge that, while these weights are subjective, they are based on our understanding of the sources and our prior coverage of software topics. Varying the weight factors allows us to emphasize different types of popularity and produce the different rankings. We then combined each weighted data source for each program and then renormalized the resulting frequency to produce an aggregate popularity score.
The Top Programming Languages was originally created by data journalist Nick Diakopoulos. Our statistical methodology advisor is Hilary Wething. Research assistance was provided by Elizabeth Wood. Rankings are computed using R.
Google
Google is the leading search engine in the world, making it an ideal fit for estimating language popularity. We measured the number of hits for each language by searching on the template, “X programming language” (with quotation marks) and manually recorded the number of results that were returned by the search. We took the measurement in July 2024. We like this measure because it indicates the volume of online information resources about each programming language.
Stack Overflow
Stack Overflow is a popular site where programmers can ask questions about coding. We recorded the number of questions tagged to each program within the last week prior to our search (August 2024). For the Mathematica/Wolfram language, we relied on the sister “Stack” for the Mathematica platform and tallied the number of programming-related questions asked in the past week. These data were gathered manually. This measure indicates what programming languages are currently trending.
IEEE Xplore Digital Library
IEEE maintains a digital library with millions of conference and journal articles covering a wide array of scientific and engineering disciplines. We searched for articles that mention each of the languages in the template “X programming” for the years 2023 and 2024, because this is the smallest timeframe for which we could access articles. For search results that returned thousands of articles, we identified the correct sample size for a 95 percent confidence interval (usually a little over 300) and pulled that number of articles. For each language we sampled, we identified the share of articles that utilize the programming language and then multiplied the total number of articles by this share to tally the likely total number of articles that reference a given programming language. We conducted this search in July 2024. This metric captures the prevalence of the different programming languages as used and referenced in engineering scholarship.
IEEE Job Site
We measured the demand for different programming languages in job postings on the IEEE Job Site. For search results that returned thousands of listings, we identified the correct sample size for a 95 percent confidence interval (usually around 300 results) and pulled that number of job listings to manually examine. For each language we sampled, we identified the share of listings that utilize the programming language and then multiplied the total number of job listings by this share to tally the likely total number of job listings that reference a given programming language. Additionally, because some of the languages we track could be ambiguous in plain text—such as lD, Go, J, Ada, and R—we searched for job postings with those words in the job description and then manually examined the results, again sampling entries if the number of results was large. The search was conducted in July 2024. We like the IEEE Job Site for its large number of non-U.S. listings, making it an ideal to measure global popularity.
CareerBuilder
We measured the demand for different programming languages on the CareerBuilder job site. We searched for “Developer” jobs offered within the United States, as this is the most popular job title for programmers. We sampled 400 job ads and manually examined them to identify which languages employers mentioned in the postings. The search was conducted in July 2024. We like the career builder site to identify the popularity of programmer jobs in the United States.
GitHub
GitHub is a public repository for many volunteer-driven open-source software projects. We used data gathered by GitHut 2.0, which measures the top 50 languages used by the number of repositories tagged with that language and draws from GitHub’s public API. We use two metrics from GitHub: repositories that have been “starred” by users to reflect long-term interests, and the number of pull requests to indicate current activity. The data cover the second quarter of 2024. These measures indicate what languages coders choose to work in when they have a personal choice.
Trinity College Dublin Library
The library of Trinity College Dublin is one of six legal deposit libraries in Ireland and the United Kingdom. A copy must be deposited with the library of any book published or distributed in Ireland, and on request any U.K. publisher or distributor must also deposit a book. We searched for all books published in the year to date that had their subject matter categorized as computer programming and totaled the number of returns. The search was conducted in June 2024. We like this library collection because it represents a large and categorized sample of works, primarily in the English language.
Discord
Discord is popular chat-room platform where many programmers exchange information. We counted the number of tags that correspond to each language. In the case of languages that could also be names of nonprogramming topics, (many nonprogramming-related topics also have dedicated Discord servers; for example, “Julia” could refer to the programming language or the Sesame Street puppet), results were manually examined. Disboard was searched in August 2024. Disboard lists many public discord servers and many young coders use the site, contributing a different demographic of coders.
-
The Top Programming Languages 2024
by Stephen Cass on 22. August 2024. at 15:03
Welcome to IEEE Spectrum’s 11th annual rankings of the most popular programming languages. As always, we combine multiple metrics from different sources to create three meta rankings. The “Spectrum” ranking is weighted towards the profile of the typical IEEE member, the “Trending” ranking seeks to spot languages that are in the zeitgeist, and the “Jobs” ranking measures what employers are looking for.
You can find a full breakdown of our methodology here, but let’s jump into our results. At the top, Python continues to cement its overall dominance, buoyed by things like popular libraries for hot fields such as AI as well as its pedagogical prominence. (For most students today, if they learn one programming language in school, it’s Python.) Python’s pretty popular with employers too, although there its lead over other general purpose languages is not as large and, like last year, it plays second fiddle to the database query language SQL, which employers like to see paired with another language. SQL popularity with employers is a natural extension of today’s emphasis on networked and cloud-based system architectures, where databases become the natural repository for all the bytes a program’s logic is chewing on.
Top Programming Languages 2024 is brought to you by the IEEE Computer Society. Get connected with the world’s largest community empowering computer science and engineering professionals.
Stalwarts like Java, Javascript, and C++ also retain high rankings, but it’s what’s going on a little further down that’s particularly interesting. Typescript—a superset of Javascript—moves up several places on all the rankings, especially for Jobs, where it climbs to fourth place, versus 11th last year. Typescript’s primary differentiator over Javascript is that it enforces static typing of variables, where the type of a variable—integer, floating point, text, and so forth—must be declared before it can be used. This allows for more error checking when Typescript programs are compiled to Javascript, and the increase in reliability has proven appealing.
Another climber is Rust, a language aimed at creating system software, like C or C++. But unlike those two languages, Rust is “memory safe”, meaning it uses a variety of techniques to ensure programs can’t write to locations in memory that they are not supposed to. Such errors are a major source of security vulnerabilities. Rust’s profile has been rising sharply, boosted by things like a February cybersecurity report from the White House calling for memory safe languages to replace C and and C++. Indeed, C’s popularity appears to be on the wane, falling from fourth to ninth place on the Spectrum ranking and from 7th to 13th on the Jobs ranking.
Two languages have entered the rankings for the first time: Apex and Solidity. Apex is designed for building business applications that use a Salesforce server as a back end, and Solidity is designed for creating smart contracts on the Ethereum blockchain.
This year also saw several languages drop out of the rankings. This doesn’t mean a language is completely dead, it just means that these languages’ signal is too weak to allow them to be meaningfully ranked. Languages that dropped out included Forth, a personal favorite of mine that’s still popular with folks building 8-bit retro systems because of its tiny footprint. A weak signal is also why we haven’t included some buzzy languages such as Zig, although those proficient in it can apparently command some high salaries.
As these other languages come and go from the rankings, I have to give the shout out to the immortals, Fortran and Cobol. Although they are around 65 years old, you can still find employers looking for programmers in both. For Fortran, this tends to be for a select group of people who are also comfortable with high-energy physics, especially the kind of high-energy physics that goes boom (and with the security clearances to match). Cobol is more broadly in demand, as many government and financial systems still rely on decades-old infrastructure—and the recent paralyzing impact of the Cloudstrike/Microsoft Windows outage incident probably hasn’t done much to encourage their replacement!
-
Meet Boardwalk Robotics’ Addition to the Humanoid Workforce
by Evan Ackerman on 22. August 2024. at 12:00
Boardwalk Robotics is announcing its entry into the increasingly crowded commercial humanoid(ish) space with Alex, a “workforce transformation” humanoid upper torso designed to work in manufacturing, logistics, and maintenance.
Before we get into Alex, let me take just a minute here to straighten out how Boardwalk Robotics is related to IHMC, the Institute for Human Machine Cognition in Pensacola, Fla. IHMC is, I think it’s fair to say, somewhat legendary when it comes to bipedal robotics—its DARPA Robotics Challenge team took second place in the final event (using a Boston Dynamics DRC Atlas), and when NASA needed someone to teach the agency’s Valkyrie humanoid to walk better, they sent it to IHMC.
Boardwalk, which was founded in 2017, has been a commercial partner with IHMC when it comes to the actual building of robots. The most visible example of this to date has been IHMC’s Nadia humanoid, a research platform which Boardwalk collaborated on and built. There’s obviously a lot of crossover between IHMC and Boardwalk in terms of institutional knowledge and experience, but Alex is a commercial robot developed entirely in-house by Boardwalk.
“We’ve used Nadia to learn a lot in the realm of dynamic locomotion research, and we’re taking all that and sticking it into a manipulation platform that’s ready for commercial work,” says Brandon Shrewsbury, Boardwalk Robotics’ CTO. “With Alex, we’re focusing on the manipulation side first, getting that well established. And then picking the mobility to match the task.”
The first thing you’ll notice about Alex is that it doesn’t have legs, at least for now. Boardwalk’s theory is that for a humanoid to be practical and cost effective in the near term, legs aren’t necessary, and that there are many tasks that offer a good return on investment where a stationary pedestal or a glorified autonomous mobile robotic base would be totally fine.
“There are going to be some problem sets that require legs, but there are many problem sets that don’t,” says Robert Griffin, a technical advisor at Boardwalk. “And there aren’t very many problem sets that don’t require halfway decent manipulation capabilities. So if we can design the manipulation well from the beginning, then we won’t have to depend on legs for making a robot that’s functionally useful.”
It certainly helps that Boardwalk isn’t at all worried about developing legs: “Every time we bring up a new humanoid, it’s something like twice as fast as the previous time,” Griffin says. This will be the eighth humanoid that IHMC has been involved in bringing up—I’d tell you more about all eight of those humanoids, but some of them are so secret that even I don’t know anything about them. Legs are definitely on the road map, but they’re not done yet, and IHMC will have a hand in their development to speed things along: It turns out that already having access to a functional (top of the line, really) locomotion stack is a big head start.
Alex’s actuators are all designed in-house, and the next version will feature new grippers that allow for quicker tool changes.Boardwalk Robotics
While the humanoid space is wide open right now and competition isn’t really an issue, looking ahead, Boardwalk sees safety as one of its primary differentiators since it’s not starting out with legs, says Shrewsbury. “For a full humanoid, there’s no way to make that completely safe. If it falls, it’s going to face-plant.” By keeping Alex on a stable base, it can work closer to humans and potentially move its arms much faster while also preserving a dynamic safety zone.
Alex is available for researchers to purchase immediately.Boardwalk Robotics
Despite its upbringing in research, Alex is not intended to be a research robot. You can buy it for research purposes, if you want, but Boardwalk will be selling Alex as a commercial robot. At the moment, Boardwalk is conducting pilot programs with Alex where they’re working in partnership with select customers, with the eventual goal of transitioning to a service model. The first few sectors that Boardwalk is targeting include logistics (because, of course) and food processing, although as Boardwalk CEO Michael Morin tells us, one of the very first pilots is (appropriately enough) in aviation.
Morin, who helped to commercialize Barrett Technologies’ WAM Arm before spending some time at Vicarious Surgical as that company went public, joined Boardwalk to help them turn good engineering into a good product, which is arguably the hardest part of making useful robots (besides all the other hardest parts). “A lot of these companies are just learning about humanoids for the first time,” says Morin. “That makes the customer journey longer. But we’re putting in the effort to educate them on how this could be implemented in their world.”
If you want an Alex of your very own, Boardwalk is currently selecting commercial partners for a few more pilots. And for researchers, the robot is available right now.
-
IBM’s Big Bet on the Quantum-Centric Supercomputer
by Jay Gambetta on 21. August 2024. at 13:00
Back in June 2022, Oak Ridge National Laboratory debuted Frontier—the world’s most powerful supercomputer. Frontier can perform a billion billion calculations per second. And yet there are computational problems that Frontier may never be able to solve in a reasonable amount of time.
Some of these problems are as simple as factoring a large number into primes. Others are among the most important facing Earth today, like quickly modeling complex molecules for drugs to treat emerging diseases, and developing more efficient materials for carbon capture or batteries.
However, in the next decade, we expect a new form of supercomputing to emerge unlike anything prior. Not only could it potentially tackle these problems, but we hope it’ll do so with a fraction of the cost, footprint, time, and energy. This new supercomputing paradigm will incorporate an entirely new computing architecture, one that mirrors the strange behavior of matter at the atomic level—quantum computing.
For decades, quantum computers have struggled to reach commercial viability. The quantum behaviors that power these computers are extremely sensitive to environmental noise, and difficult to scale to large enough machines to do useful calculations. But several key advances have been made in the last decade, with improvements in hardware as well as theoretical advances in how to handle noise. These advances have allowed quantum computers to finally reach a performance level where their classical counterparts are struggling to keep up, at least for some specific calculations.
For the first time, we here at IBM can see a path toward useful quantum computers, and we can begin imagining what the future of computing will look like. We don’t expect quantum computing to replace classical computing. Instead, quantum computers and classical computers will work together to run computations beyond what’s possible on either alone. Several supercomputer facilities around the world are already planning to incorporate quantum-computing hardware into their systems, including Germany’s Jupiter, Japan’s Fugaku, and Poland’s PSNC. While it has previously been called hybrid quantum-classical computing, and may go by other names, we call this vision quantum-centric supercomputing.
A Tale of Bits and Qubits
At the heart of our vision for a quantum-centric supercomputer is the quantum hardware, which we call a quantum processing unit (QPU). The power of the QPU to perform better than classical processing units in certain tasks comes from an operating principle that’s fundamentally different, one rooted in the physics of quantum mechanics.
In the standard or “classical” model of computation, we can reduce all information to strings of binary digits, bits for short, which can take on values of either 0 or 1. We can process that information using simple logic gates, like AND, OR, NOT, and NAND, which act on one or two bits at a time. The “state” of a classical computer is determined by the states of all its bits. So, if you have N bits, then the computer can be in just one of 2N states.
But a quantum computer has access to a much richer repertoire of states during computation. A quantum computer also has bits. But instead of just 0 and 1, its quantum bits— qubits—via a quantum property known as superposition, represent 0, 1, or a linear combination of both. While a digital computer can be in just one of those 2N states, a quantum computer can be in many logical states at once during the computation. And the superpositions the different qubits are in can be correlated with one another in a fundamental way, thanks to another quantum property known as entanglement. At the end of the computation, the qubit assumes just one state, chosen based on probabilities generated during the running of the quantum algorithm.
It’s not obvious how this computing paradigm can outperform the classical one. But in 1994, Peter Shor, a mathematician at MIT, discovered an algorithm that, using the quantum-computing paradigm, could divide large numbers into their prime factors exponentially faster than the best classical algorithm. Two years later, Lov Grover discovered a quantum algorithm that could find a particular entry in a database much faster than a classical one could.
Perhaps most importantly, since quantum computers follow the laws of quantum mechanics, they are the right tool for simulating the fundamentally quantum phenomena of our world, such as molecular interactions for drug discovery or materials design.
The Quantum-Centric Supercomputer’s Center
Before we can build a quantum-centric supercomputer, we have to make sure it’s capable of doing something useful. Building a capable enough QPU relies on constructing hardware that can re-create counterintuitive quantum behaviors.
Here at IBM, the basic building block of a quantum computation—the qubit—is made out of superconducting components. Each physical qubit consists of two superconducting plates, which act as a capacitor, wired to components called Josephson junctions, which act as a special lossless, nonlinear inductor.
The current flowing across Josephson junctions is quantized—fixed to discrete values. The Josephson junctions ensure that only two of those values (or their superpositions) are realistically accessible. The qubit is encoded in two current levels, one representing a 0, the other a 1. But, as mentioned, the qubit can also exist in a superposition of the 0 and 1 states.
Because superconductors need frigid temperatures to maintain superconductivity, the qubits and some of their control circuitry are held inside a specialty liquid-helium fridge called a dilution refrigerator.
We change the qubit states and couple qubits together with quantum instructions, commonly known as gates. These are a series of specially crafted microwave waveforms. A QPU includes all of the hardware responsible for accepting a set of quantum instructions—called a quantum circuit—and returning a single output represented by a binary string. The QPU includes the qubits plus components that amplify signals, the control electronics, and the classical computation required for tasks such as holding the instructions in memory, accumulating and separating signals from noise, and creating single binary outputs. We etch components like qubits, resonators for readouts, output filters, and quantum buses into a superconducting layer deposited on top of a silicon chip.
But it’s a challenge trying to control qubits at the supersensitive quantum level. External noise, noise from the electronics, and cross talk between control signals for different qubits all destroy the fragile quantum properties of the qubits. Controlling these noise sources has been key in reaching the point where we can envision useful quantum-centric supercomputers.
Getting the Quantum Stuff up to Snuff
No one has yet conclusively demonstrated quantum advantage—that is, a quantum computer that outperforms the best classical one on a real-world relevant task. Demonstrating true quantum advantage would herald a new era of computing, where previously intractable tasks would now be within reach.
Before we can approach this grandiose goal, we have to set our sights a bit lower, to a target we call quantum utility. Quantum utility is the ability of quantum hardware to outperform brute-force classical calculations of a quantum circuit. In other words, it’s the point where quantum hardware is better at doing quantum computations than a traditional computer is.
This may sound underwhelming, but it is a necessary stepping-stone on the way to quantum advantage. In recent years, the quantum community has finally reached this threshold. Demonstrating quantum utility of our QPU, which we did in 2023, has convinced us that our quantum hardware is advanced enough to merit being built into a quantum-centric supercomputer. Achieving this milestone has taken a combination of advances, including both hardware and algorithmic improvements.
Since 2019, we’ve been incorporating advances in semiconductor fabrication to introduce 3D integration to our chips. This gave us access to qubits from a controller chip placed below the qubit plane to reduce the wiring on the chip, a potential source of noise. We also introduced readout multiplexing, which allows us to access the information from several qubits with a single wire, drastically reducing the amount of hardware we have to put in the dilution refrigerator.
In 2023, we implemented a new way to perform quantum gates—the steps of a program that change the value of the qubits—on our hardware, using components called tunable couplers. Previously, we prevented cross talk by fabricating the qubits that respond to different frequencies so that they wouldn’t react to microwave pulses meant for other qubits. But this made it too difficult for the qubits to perform the essential task of talking to one another, and it also made the processors slow. With tunable couplers, we don’t need the frequency-specific fabrication. Instead, we introduced a sort of “on-off” switch, using magnetic fields to decide whether or not a qubit should talk to another qubit. The result: We virtually eliminated cross-talk errors between qubits, allowing us to run much faster, more reliable gates.
As our hardware improved, we also demonstrated that we could deal with some noise using an error mitigation algorithm. Error mitigation can be done in many ways. In our case, we run quantum programs, analyze how the noise in our system changes the program outputs, and then create a noise model. Then we can use classical computing and our noise model to recover what a noise-free result would look like. The surrounding hardware and software of our quantum computer therefore includes classical computing capable of performing error mitigation, suppression, and eventually, error correction.
Alongside ever-improving hardware advances, we teamed up with the University of California, Berkeley, to demonstrate in 2023 that a quantum computer running our 127-qubit quantum chip, Eagle, could run circuits beyond the ability of brute-force classical simulation—that is, methods where the classical computer exactly simulates the quantum computer in order to run the circuit, reaching quantum utility. And we did so for a real condensed-matter physics problem—namely, finding the value of a property called magnetization for a system of simplified atoms with a structure that looked like the layout of our processors’ qubits.
Left: A quantum processing unit is more than just a chip. It includes the interconnects, amplifiers, and signal filtering. It also requires the classical hardware, including the room-temperature classical computers needed to receive and apply instructions and return outputs. Right: At the heart of an IBM quantum computer is a multilayer semiconductor chip etched with superconducting circuits. These circuits comprise the qubits used to perform calculations. Chips are divided into a layer with the qubits, a layer with resonators for readout, and multiple layers of wiring for input and output.
Error Correction to the Rescue
We were able to demonstrate the ability of our quantum hardware outperforming brute-force classical simulation without leveraging the most powerful area of quantum-computing theory: quantum error correction.
Unlike error mitigation, which deals with noise after a computation, quantum error correction can remove noise as it arises during the process. And it works for a more general kind of noise; you don’t need to figure out a specific noise model first. Plus, while error mitigation is limited in its ability to scale as the complexity of quantum circuits grows, error correction will continue to work at large scales.
Error Correction
But quantum error correction comes at a huge cost: It requires more qubits, more connectivity, and more gates. For every qubit you want to compute with, you may need many more to enable error correction. Recent advances in improving hardware and finding better error-correcting codes have allowed us to envision an error-corrected supercomputer that can make those costs worthwhile.
Quantum error-correcting schemes are a bit more involved than error correction in traditional binary computers. To work at all, these quantum schemes require that the hardware error rate is below a certain threshold. Since quantum error correction’s inception, theorists have devised new codes with more relaxed thresholds, while quantum-computer engineers have developed better-performing systems. But there hasn’t yet been a quantum computer capable of using error correction to perform large-scale calculations.
Meanwhile, error-correction theory has continued to advance. One promising finding by Moscow State University physicists Pavel Panteleev and Gleb Kalachev inspired us to pursue a new kind of error-correcting code for our systems. Their 2021 paper demonstrated the theoretical existence of “good codes,” codes where the number of extra qubits required to perform error correction scales more favorably.
This led to an explosion of research into a family of codes called quantum low-density parity check codes, or qLDPC codes. Earlier this year, our team published a qLDPC code with an error threshold high enough that we could conceivably implement it on near-term quantum computers; the amount of required connectivity between qubits was only slightly beyond what our hardware already supplies. This code would need only a tenth the number of qubits as previous methods to achieve error correction at the same level.
These theoretical developments allow us to envision an error-corrected quantum computer at experimentally accessible scales, provided we can connect enough quantum processing power together, and leverage classical computing as much as possible.
Hybrid Classical-Quantum Computers for the Win
To take advantage of error correction, and to reach large enough scales to solve human-relevant problems with quantum computers, we need to build larger QPUs or connect multiple QPUs together. We also need to incorporate classical computing with the quantum system.
Quantum-centric supercomputers will include thousands of error-corrected qubits to unlock the full power of quantum computers. Here’s how we’ll get there.
2024
Heron
→ 156 qubits
→ 5K gates before errors set in
2025
Flamingo
→ Introduce l-couplers between chips
→ Connect 7 chips for 7 x 156 = 1,092 qubits
→ 5K gates before errors set in
2027
Flamingo
→ l-couplers between chips
→ 7 x 156 = 1,092 qubits
→ Improved hardware and error mitigation
→ 10K gates before errors set in
2029
Starling
→ 200 qubits
→ l-, m-, and c-couplers combined
→ Error correction
→ 100M gates
2030
BlueJay
→ 2,000 qubits
→ Error correction
→ 1B gates
Last year, we released a machine we call the IBM Quantum System Two, which we can use to start prototyping error mitigation and error correction in a scalable quantum computing system. System Two relies on larger, modular cryostats, allowing us to place multiple quantum processors into a single refrigerator with short-range interconnects, and then combine multiple fridges into a bigger system, kind of like adding more racks to a traditional supercomputer.
Along with the System Two release, we also detailed a 10-year plan for realizing our vision. Much of the early hardware work on that road map has to do with interconnects. We’re still developing the interconnects required to connect quantum chips into larger chips like Lego blocks, which we call m-couplers. We’re also developing interconnects to transfer quantum information between more distant chips, called l-couplers. We hope to prototype both m- and l-couplers by the end of this year. We’re also developing on-chip couplers that link qubits on the same chip that are more distant than their nearest neighbors—a requirement of our newly developed error-correction code. We plan to deliver this c-coupler by the end of 2026. In the meantime, we’ll be improving error mitigation so that by 2028, we can run a quantum program across seven parallel quantum chips, each chip capable of performing up to 15,000 accurate gates before the errors set in, on 156 qubits.
We’re also continuing to advance error correction. Our theorists are always looking for codes that require fewer extra qubits for more error-correcting power and allow for higher error thresholds. We must also determine the best way to run operations on information that’s encoded into the error-correcting code, and then decode that information in real time. We hope to demonstrate those by the end of 2028. That way, in 2029, we can debut our first quantum computer incorporating both error mitigation and error correction that can run up to 100 million gates until the errors take hold, on 200 qubits. Further advances in error correction will allow us to run a billion gates on 2,000 qubits by 2033.
Knitting Together a Quantum-Centric Supercomputer
The ability to mitigate and correct errors removes a major roadblock in the way of full-scale quantum computing. But we still don’t think it’ll be enough to tackle the largest, most valuable problems. For that reason, we’ve also introduced a new way of running algorithms, where multiple quantum circuits and distributed classical computing are woven together into a quantum-centric supercomputer.
Many envision the “quantum computer” as a single QPU, working on its own to run programs with billions of operations on millions of physical qubits. Instead, we envision computers incorporating multiple QPUs, running quantum circuits in parallel with distributed classical computers.
Combining the strengths of quantum and classical
Quantum-centric supercomputing leverages quantum and classical resources in parallelized workloads to run computations larger than what was possible before. A quantum-centric supercomputer is a system optimized to orchestrate work across the quantum computers and advanced classical compute clusters in the same data center.
Recent work has demonstrated techniques that let us run quantum circuits much more efficiently by incorporating classical computing with quantum processing. These techniques, called circuit knitting, break down a single quantum-computing problem into multiple quantum-computing problems and then run them in parallel on quantum processors. And then a combination of quantum and classical computers knit the circuit results together for the final answer.
Another technique uses the classical computer to run all but the core, intrinsically quantum part of the calculation. It is this last vision that we believe will realize quantum advantage first.
Therefore, a quantum computer doesn’t just include one quantum processor, its control electronics, and its dilution refrigerator—it also includes the classical processing required to perform error correction, and error mitigation.
We haven’t realized a fully integrated quantum-centric supercomputer yet. But we’re laying the groundwork with System Two, and Qiskit, our full-stack quantum-computing software for running large quantum workloads. We are building middleware capable of managing circuit knitting, and of provisioning the appropriate computing resources when and where they’re required. The next step is to mature our hardware and software infrastructure so that quantum and classical can extend one another to do things beyond the capabilities of either.
Today’s quantum computers are now scientific tools capable of running programs beyond the brute-force ability of classical simulation, at least when simulating certain quantum systems. But we must continue improving both our quantum and classical infrastructure so that, combined, it’s capable of speeding up solutions for problems relevant to humanity. With that in mind, we hope that the broader computing community will continue researching new algorithms incorporating circuit knitting, parallelized quantum circuits, and error mitigation in order to find use cases that can benefit from quantum in the near term.
And we look forward to a day when the Top 500 list of most powerful supercomputers will include machines that have quantum processors at their hearts.
-
Ansys SimAI Software Predicts Fully Transient Vehicle Crash Outcomes
by Ansys on 20. August 2024. at 18:09
The Ansys SimAI™ cloud-enabled generative artificial intelligence (AI) platform combines the predictive accuracy of Ansys simulation with the speed of generative AI. Because of the software’s versatile underlying neural networks, it can extend to many types of simulation, including structural applications.
This white paper shows how the SimAI cloud-based software applies to highly nonlinear, transient structural simulations, such as automobile crashes, and includes:- Vehicle kinematics and deformation
- Forces acting upon the vehicle
- How it interacts with its environment
- How understanding the changing and rapid sequence of events helps predict outcomes
These simulations can reduce the potential for occupant injuries and the severity of vehicle damage and help understand the crash’s overall dynamics. Ultimately, this leads to safer automotive design.
-
IEEE Introduces Digital Certificates Documenting Volunteer Roles
by Joanna Goodrich on 20. August 2024. at 18:00
IEEE Collabratec has made it easier for volunteers to display their IEEE positions. The online networking platform released a new benefit this year for its users: digital certificates for IEEE volunteering. They reflect contributions made to the organization, such as leading a committee or organizing an event.
Members can download the certificates and add them to their LinkedIn profile or résumé. Volunteers also can print their certificates to frame and display in their office.
Each individualized document includes the person’s name, the position they’ve held, and the years served. Every position held has its own certificate. The member’s list of roles is updated annually.
The feature is a result of a top recommendation to improve volunteer recognition made by delegates at the 2023 IEEE Sections Congress, according to Deepak Mathur. The senior member is vice president of IEEE Member and Geographic Activities. The new feature “respects the time and effort of our volunteers and is a testament to the power and versatility of the Collabratec platform,” Mathur said in an announcement.
Members can download their certificates by selecting the Certificates tab on their Collabratec page and scrolling to each of their positions.
To learn more about IEEE Collabratec, check out the user guide, FAQs, and users’ forum.
-
Nasir Ahmed: An Unsung Hero of Digital Media
by Willie D. Jones on 19. August 2024. at 12:00
Stop for a second and think about the Internet without digital images or video. There would be no faces on Facebook. Instagram and TikTok probably wouldn’t exist. Those Zoom meetings that took the place of in-person gatherings for school or work during the height of the COVID-19 pandemic? Not an option.
Digital audio’s place in our Internet-connected world is just as important as still images and video. It has changed the music business—from production to distribution to the way fans buy, collect, and store their favorite songs.
What do those millions of profiles on LinkedIn, dating apps, and social media platforms (and the inexhaustible selection of music available for download online) have in common? They rely on a compression algorithm called the discrete cosine transform, or DCT, which played a major role in allowing digital files to be transmitted across computer networks.
“DCT has been one of the key components of many past image- and video-coding algorithms for more than three decades,” says Touradj Ebrahimi, a professor at Ecole Polytechnique Fédérale de Lausanne, in Switzerland, who currently serves as chairman of the JPEG standardization committee. “Only a few image-compression standards not using DCT exist today,” he adds.
The Internet applications people use every day but largely take for granted were made possible by scientists and engineers who, for the most part, toiled in anonymity. One such “hidden figure” is Nasir Ahmed, the Indian-American engineer who figured out an elegant way to cut down the size of digital image files without sacrificing their most critical visual details.
Ahmed published his seminal paper about the discrete cosine transform compression algorithm he invented in 1974, a time when the fledgling Internet was exclusively dial-up and text-based. There were no pictures accompanying the words, nor could there have been, because Internet data was transmitted over standard copper telephone landlines, which was a major limitation on speed and bandwidth.
“Only a few image-compression standards not using DCT exist today.” –Touradj Ebrahimi, EPFL
These days, with the benefit of superfast chips and optical-fiber networks, data download speeds for a laptop with a fiber connection reach 1 gigabit per second. So, a music lover can download a 4-minute song to their laptop (or more likely a smartphone) in a second or two. In the dial-up era, when Internet users’ download speeds topped out at 56 kilobits per second (and were usually only half that fast), pulling down the same song from a server would have taken nearly all day. Getting a picture to appear on a computer’s screen was a process akin to watching grass grow.
Ahmed was convinced there had to be a way to cut down the size of digital files and speed up the process. He set off on a quest to represent with ones and zeros what is critical to an image being legible, while tossing aside the bits that are less important. The answer, which built on the earlier work of mathematician and information-theory pioneer Claude Shannon, took a while to come into focus. But because of Ahmed’s determination and unwavering belief in the value of what he was doing, he persevered even after others told him that it was not worth the effort.
Raised to Love Technology
It seemed almost preordained that Ahmed would have a career in one of the STEM fields. Nasir, who was born in Bengaluru, India, in 1940, was raised by his maternal grandparents. Ahmed’s grandfather was an electrical engineer who told him that he had been sent to the United States in 1919 to work at General Electric‘s location in Schenectady, N.Y. He shared tales of his time in the United States with his grandson and encouraged young Nasir to emigrate there. In 1961, after earning a bachelor’s degree in electrical engineering at the University of Visvesvaraya College of Engineering, in Bengaluru, Ahmed did just that, leaving India that fall for graduate school at the University of New Mexico, in Albuquerque. Ahmed earned a master’s degree and a Ph.D. in electrical engineering in 1963 and 1966, respectively.
During his first year in Albuquerque, he met Esther Parente, a graduate student from Argentina. They soon became inseparable and were married while he was working toward his doctorate. Sixty years later, they are still together.
The Seed of an Idea
In 1966, Ahmed, fresh out of grad school with his Ph.D., was hired as a principal research engineer at Honeywell’s newly created computer division. While there, Ahmed was first exposed to Walsh functions, a technique for analyzing digital representations of analog signals. The fast algorithms that could be created based on Walsh functions had many potential applications. Ahmed focused on using these signal-processing and analysis techniques to reduce the file size of a digital image without losing too much of the visual detail in the uncompressed version.
That research focus remained his primary interest when he returned to academia, taking a job as a professor in the electrical and computer engineering department at Kansas State University, in 1968.
Ahmed, like dozens of other researchers around the globe, was obsessed with finding the answer to a single question: How do you create a mathematical formula for deciphering which of the ones and zeros that represent a digital image need to be kept and which can be thrown away? The things he’d learned at Honeywell gave him a framework for understanding the elements of the problem and how to attack it. But the majority of the credit for the eventual breakthrough has to go to Ahmed’s steely determination and willingness to take a gamble on himself.
In 1972, he sought grant funding that would let him afford to spend the months between Kansas State’s spring and fall semesters furthering his ideas. He applied for a U.S. National Science Foundation grant, but was denied. Ahmed recalls the moment: “I had a strong intuition that I could find an efficient way to compress digital signal data. But to my surprise, the reviewers said the idea was too simple, so they rejected the proposal.”
Undaunted, Ahmed and his wife worked to make the salary he earned during the nine-month school year last through the summer so he could focus on his research. Money was tight, the couple recalls, but that moment of financial belt-tightening only seemed to heighten Ahmed’s industriousness. They persevered, and Ahmed’s long days and late nights in the lab eventually yielded the desired result.
DCT Compression Comes Together
Ahmed took a technique for turning the array of image-processing data representing an image’s pixels into a waveform, effectively rendering it as a series of waves with oscillating frequencies, and combined it with cosine functions that were already being used to model phenomena such as light waves, sound waves, and electric current. The result was a long string of numbers with values bounded by 1 and –1. Ahmed realized that by quantizing this string of values and performing a Fourier transformation to break the function into its constituent frequencies, each pixel’s data could be represented in a way that was helpful for deciding what data points must be kept and what could be omitted. Ahmed observed that the lower-frequency waves corresponded to the necessary or “high information” regions of the image, while the higher-frequency waves represented the bits that were less important and could therefore be approximated. The compressed-image files he and his team produced were one-tenth the size of the originals. What’s more, the process could be reversed, and a shrunken data file would yield an image that was sufficiently similar to the original.
After another two years of laborious testing, with he and his two collaborators running computer programs written on decks of data punch cards, the trio published a paper in IEEE Transactions On Computers titled “Discrete Cosine Transform” in January 1974. Though the paper’s publication did not make it immediately clear, the worldwide search for a reliable method of doing the lossy compression that Claude Shannon had postulated in the 1940s was over.
JPEGs, MPEGs, and More
It wasn’t until 1983 that the International Organization for Standardization (ISO) began working on the technology that would allow photo-quality images to accompany text on the screens of computer terminals. To that end, ISO established the Joint Photographic Experts Group, better known by the ubiquitous acronym JPEG. By the time the first JPEG standard was published in 1992, DCT and advances made by a cadre of other researchers had come to be recognized by the group as basic elements of their method for the digital compression and coding of still images. “This is the beauty of standardization, where several dozen bright minds are behind the success of advances such as JPEG,” says Ebrahimi.
And because video can be described as a succession of still images, Ahmed’s technique was also well suited to making video files smaller. DCT was the compression technique of choice when ISO and the international Electrotechnical Commission (IEC) established the Moving Picture Experts Group, or MPEG, for the compression and coding of audio, video, graphics, and genomic data in 1988. When the first MPEG standard was published in 1993, the World Wide Web that now includes Google Maps, dating apps, and e-commerce businesses was just four years old.
The ramping up of computer speeds and network bandwidth during that decade—along with the ability to transmit pictures and video via much smaller files—quickly transformed the Internet before anyone knew that Amazon would eventually let readers judge millions of books by their covers.
Having solved the problem that had monopolized his time and attention for several years, Ahmed resumed his career in academia. In 1993, the year the first MPEG standard went on the books, Ahmed left Kansas State and returned to the University of New Mexico. There he was a presidential professor of electrical and computer engineering until 1989, when he was promoted to chair of the ECE department. Five years after that, he became dean of UNM’s school of engineering. Ahmed held that post for two years until he was named associate provost for research and dean of graduate studies. He stayed in that job until he retired from the university in 2001 and was named professor emeritus.
-
Taiwan Reboots Its Solar-Power Fishponds
by Peter Fairley on 19. August 2024. at 12:00
A maze of brackish and freshwater ponds covers Taiwan’s coastal plain, supporting aquaculture operations that produce roughly NT $30 billion (US $920 million) worth of seafood every year. Taiwan’s government is hoping that the more than 400 square kilometers of fishponds can simultaneously produce a second harvest: solar power.
What is aquavoltaics?
That’s the impetus behind the new 42.9-megawatt aquavoltaics facility in the southern city of Tainan. To build it, Taipei-based Hongde Renewable Energy bought 57.6 hectares of abandoned land in Tainan’s fishpond-rich Qigu district, created earthen berms to delineate the two dozen ponds, and installed solar panels along the berms and over six reservoir ponds.
Tony Chang, general manager of the Hongde subsidiary Star Aquaculture, says 18 of the ponds are stocked with mullet (prized for their roe) and shrimp, while milkfish help clean the water in the reservoir ponds. In 2023, the first full year of operation, Chang says his team harvested over 100,000 kilograms of seafood. This August, they began stocking a cavernous indoor facility, also festooned with photovoltaics, to cultivate white-legged shrimp.
A number of other countries have been experimenting with aquavoltaics, including China, Chile, Bangladesh, and Norway, extending the concept to large solar arrays floating on rivers and bays. But nowhere else is the pairing of aquaculture and solar power seen as so crucial to the economy. Taiwan is striving to massively expand renewable generation to sustain its semiconductor fabs, and solar is expected to play a large role. But on this densely populated island—slightly larger than Maryland, smaller than the Netherlands—there’s not a lot of open space to install solar panels. The fishponds are hard to ignore. By the end of 2025, the government is looking to install 4.4 gigawatts of aquavoltaics to help meet its goal of 20 GW of solar generation.
Is Taiwan’s aquavoltaics plan unrealistic?
Meanwhile, though, solar developers are struggling to deliver on Taiwan’s ambitious goals, even as some projections suggest Taiwan will need over eight times more solar by 2050. And aquavoltaics in particular have come under scrutiny from environmental groups. In 2020, for example, reporter Cai Jiashan visited 100 solar plants built on agricultural land, including fishponds, and found dozens of cases where solar developers built more solar capacity than the law intended, or secured permits based on promises of continued farming that weren’t kept.
Star Aquaculture grows milkfish to help clean water for its breeding ponds.HDRenewables
On 7 July 2020, Taiwan’s Ministry of Agriculture responded by restricting solar development on farmland, in what the solar industry called the “Double-Seven Incident.” Many aquavoltaic projects were canceled while others were delayed. The latter included a 10-MW facility in Tainan that Google had announced to great fanfare in 2019 as its first renewable-energy investment in Asia, to supply power for the company’s Taiwan data centers. The array finally started up in 2023, three years behind schedule.
Critics of Taiwan’s renewed aquavoltaic plans thus see the government’s goal as unrealistic. Yuping Chen, executive director of the Taiwan Environment and Planning Association, a Taipei-based nonprofit dedicated to resolving conflicts between solar energy and agriculture, says of aquavoltaics, “It is claimed to be crucial by the government, but it’s impossible to realize.”
How aquavoltaics could revive fishing, boost revenue
Solar developers and government officials who endorse aquavoltaics argue that such projects could revive the island’s traditional fishing community. Taiwan’s fishing villages are aging and shrinking as younger people take city jobs. Climate change has also taken a toll. Severe storms damage fishpond embankments, while extreme heat and rainfall stress the fish.
4.4
Gigawatts of aquavoltaics that Taiwan wants to install by the end of 2025
Solar development could help reverse these trends. Several recent studies examining fishponds in Taiwan found that adding solar improves profitability, providing an opportunity to reinvigorate communities if agrivoltaic investors share their returns. Alan Wu, deputy director of the Green Energy Initiative at Taiwan’s Industrial Technology Research Institute, says the Hsinchu-based lab has opened a research station in Tainan to connect solar and aquaculture firms. ITRI is helping aquavoltaics facilities boost their revenues by figuring out how they can raise “species of high economic value that are normally more difficult to raise,” Wu says.
Such high-value products include the 27,000 pieces of sun-dried mullet roe that Hongde Renewable Energy’s Tainan site produced last year. The new indoor facility, meanwhile, should boost yields of the relatively pricey whiteleg shrimp. Chang expects the indoor harvests to fetch $500,000 to $600,000 annually, compared to $800,000 to $900,000 from the larger outdoor ponds.
The solar roof over the 100,000-liter indoor growth tanks protects the 2.7 million shrimp against weather and bird droppings. Chang says a patent-pending drain mechanically removes waste from each tank, and also sucks out the shrimp when they’re ready for harvest.
Land that Star Aquaculture set aside for wildlife now attracts endangered birds like the black-faced spoonbill [left] and the oriental stork [right].iStock (2)
The company has also set aside 9 percent of the site for wildlife, in response to concerns from conservationists. “Egrets, endangered oriental storks, and black-faced spoonbills continue to use the site,” Chang says. “If it was all covered with PV, it could impact their habitat.”
Such measures may not satisfy environmentalists, though. In a review published last month, researchers at Fudan University in Shanghai and two Chinese power firms concluded that China’s floating aquavoltaic installations—some of which already span 5 square kilometers—will “inevitably” alter the marine environment.
Aquavoltaic facilities that are entirely indoors may be an even harder sell as they scale up. Toshiba is backing such a plant in Tainan, to generate 120 MW for an unspecified “semiconductor manufacturer,” with plans for a 360-MW expansion. The resulting buildings could exclude wildlife from 5 square kilometers of habitat. Indoor projects could compensate by protecting land elsewhere. But, as Chen of the Taiwan Environment and Planning Association notes, developers of such sites may not take such measures unless they’re required by law to do so.
-
Electrostatic Motors Reach the Macro Scale
by Glenn Zorpette on 18. August 2024. at 12:00
It’s a pretty sure bet that you couldn’t get through a typical day without the direct support of dozens of electric motors. They’re in all of your appliances not powered by a hand crank, in the climate-control systems that keep you comfortable, and in the pumps, fans, and window controls of your car. And although there are many different kinds of electric motors, every single one of them, from the 200-kilowatt traction motor in your electric vehicle to the stepper motor in your quartz wristwatch, exploits the exact same physical phenomenon: electromagnetism.
For decades, however, engineers have been tantalized by the virtues of motors based on an entirely different principle: electrostatics. In some applications, these motors could offer an overall boost in efficiency ranging from 30 percent to close to 100 percent, according to experiment-based analysis. And, perhaps even better, they would use only cheap, plentiful materials, rather than the rare-earth elements, special steel alloys, and copious quantities of copper found in conventional motors.
“Electrification has its sustainability challenges,” notes Daniel Ludois, a professor of electrical engineering at the University of Wisconsin in Madison. But “an electrostatic motor doesn’t need windings, doesn’t need magnets, and it doesn’t need any of the critical materials that a conventional machine needs.”
Such advantages prompted Ludois to cofound a company, C-Motive Technologies, to build macro-scale electrostatic motors. “We make our machines out of aluminum and plastic or fiberglass,” he says. Their current prototype is capable of delivering torque as high as 18 newton meters and power at 360 watts (0.5 horsepower)—characteristics they claim are “the highest torque and power measurements for any rotating electrostatic machine.”
The results are reported in a paper, “Synchronous Electrostatic Machines for Direct Drive Industrial Applications,” to be presented at the 2024 IEEE Energy Conversion Congress and Exposition, which will be held from 20 to 24 October in Phoenix, Ariz. In the paper, Ludois and four colleagues describe an electrostatic machine they built, which they describe as the first such machine capable of “driving a load performing industrial work, in this case, a constant-pressure pump system.”
Making Electrostatic Motors Bigger
The machine, which is hundreds of times more powerful than any previous electrostatic motor, is “competitive with or superior to air-cooled magnetic machinery at the fractional [horsepower] scale,” the authors add. The global market for fractional horsepower motors is more than US $8.7 billion, according to consultancy Business Research Insights.
C-Motive’s 360-watt motor has a half dozen each of rotors and stators, shown in yellow in this cutaway illustration.C-Motive Technologies
Achieving macro scale wasn’t easy. Electrostatic motors have been available for years, but today, these are tiny units with power output measured in milliwatts. “Electrostatic motors are amazing once you get below about the millimeter scale, and they get better and better as they get smaller and smaller,” says Philip Krein, a professor of electrical engineering at the University of Illinois Urbana-Champaign. “There’s a crossover at which they are better than magnetic motors.” (Krein does not have any financial connection to C-Motive.)
For larger motors, however, the opposite is true. “At macro scale, electromagnetism wins, is the textbook answer,” notes Ludois. “Well, we’ve decided to challenge that wisdom.”
For this quest he and his team found inspiration in a lesser-known accomplishment of one of the United States’ founding fathers. “The fact is that Benjamin Franklin built and demonstrated a macroscopic electrostatic motor in 1747,” says Krein. “He actually used the motor as a rotisserie to grill a turkey on a riverbank in Philadelphia” (a fact unearthed by the late historian I. Bernard Cohen for his 1990 book Benjamin Franklin’s Science ).
Krein explains that the fundamental challenge in attempting to scale electrostatic motors to the macro world is energy density. “The energy density you can get in air at a reasonable scale with an electric-field system is much, much lower—many orders of magnitude lower—than the density you can get with an electromagnetic system.” Here the phrase “in air” refers to the volume within the motor, called the “air gap,” where the machine’s fields (magnetic for the conventional motor, electric for the electrostatic one) are deployed. It straddles the machine’s key components: the rotor and the stator.
Let’s unpack that. A conventional electric motor works because a rotating magnetic field, set up in a fixed structure called a stator, engages with the magnetic field of another structure called a rotor, causing that rotor to spin. The force involved is called the Lorentz force. But what makes an electrostatic machine go ‘round is an entirely different force, called the Coulomb force. This is the attractive or repulsive physical force between opposite or like electrical charges.
Overcoming the Air Gap Problem
C-Motive’s motor uses nonconductive rotor and stator disks on which have been deposited many thin, closely spaced conductors radiating outward from the disk’s center, like spokes in a bicycle wheel. Precisely timed electrostatic charges applied to these “spokes” create two waves of voltage, one in the stator and another in the rotor. The phase difference between the rotor and stator waves is timed and controlled to maximize the torque in the rotor caused by this sequence of attraction and repulsion among the spokes. To further wring as much torque as possible, the machine has half a dozen each of rotors and stators, alternating and stacked like compact discs on a spindle.
The 360-watt motor is hundreds of times more powerful than previous electrostatic motors, which have power output generally measured in milliwatts.C-Motive Technologies
The machine would be feeble if the dielectric between the charges was air. As a dielectric, air has low permittivity, meaning that an electric field in air can not store much energy. Air also has a relatively low breakdown field strength, meaning that air can support only a fairly weak electric field before it breaks down and conducts current in a blazing arc. So one of the team’s greatest challenges was producing a dielectric fluid that has a much higher permittivity and breakdown field strength than air, and that was also environmentally friendly and nontoxic. To minimize friction, this fluid also had to have very low viscosity, because the rotors would be spinning in it. A dielectric with high permittivity concentrates the electric field between oppositely charged electrodes, enabling greater energy to be stored in the space between them. After screening hundreds of candidates over several years, the C-Motive team succeeded in producing an organic liquid dielectric with low viscosity and a relative permittivity in the low 20s. For comparison, the relative permittivity of air is 1.
Another challenge was supplying the 2,000 volts their machine needs to operate. High voltages are necessary to create the intense electric fields between the rotors and stators. To precisely control these fields, C-Motive was able to take advantage of the availability of inexpensive and stupendously capable power electronics, according to Ludois. For their most recent motor, they developed a drive system based on readily available 4.5-kilovolt insulated-gate bipolar transistors, but the rate of advancement in power semiconductors means they have many attractive choices here, and will have even more in the near future.
Ludois reports that C-Motive is now testing a 750-watt (1 hp) motor in applications with potential customers. Their next machines will be in the range of 750 to 3,750 watts (1 to 5 hp), he adds. These will be powerful enough for an expanded range of applications in industrial automation, manufacturing, and heating, ventilating, and air conditioning.
It’s been a gratifying ride for Ludois. “For me, a point of creative pride is that my team and I are working on something radically different that, I hope, over the long term, will open up other avenues for other folks to contribute.”
-
Video Friday: Silly Robot Dog Jump
by Evan Ackerman on 16. August 2024. at 16:45
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.
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!
The title of this video is “Silly Robot Dog Jump” and that’s probably more than you need to know.
[ Deep Robotics ]
It’ll be great when robots are reliably autonomous, but until they get there, collaborative capabilities are a must.
[ Robust AI ]
I am so INCREDIBLY EXCITED for this.
[ IIT Instituto Italiano di Tecnologia ]
In this 3 minutes long one-take video, the LimX Dynamics CL-1 takes on the challenge of continuous heavy objects loading among shelves in a simulated warehouse, showcasing the advantages of the general-purpose form factor of humanoid robots.
[ LimX Dynamics ]
Birds, bats and many insects can tuck their wings against their bodies when at rest and deploy them to power flight. Whereas birds and bats use well-developed pectoral and wing muscles, how insects control their wing deployment and retraction remains unclear because this varies among insect species. Here we demonstrate that rhinoceros beetles can effortlessly deploy their hindwings without necessitating muscular activity. We validated the hypothesis using a flapping microrobot that passively deployed its wings for stable, controlled flight and retracted them neatly upon landing, demonstrating a simple, yet effective, approach to the design of insect-like flying micromachines.
[ Nature ]
Agility Robotics’ CTO, Pras Velagapudi, talks about data collection, and specifically about the different kinds we collect from our real-world robot deployments and generally what that data is used for.
[ Agility Robotics ]
Robots that try really hard but are bad at things are utterly charming.
[ University of Tokyo JSK Lab ]
The DARPA Triage Challenge unsurprisingly has a bunch of robots in it.
[ DARPA ]
The Cobalt security robot has been around for a while, but I have to say, the design really holds up—it’s a good looking robot.
[ Cobalt AI ]
All robots that enter elevators should be programmed to gently sway back and forth to the elevator music. Even if there’s no elevator music.
[ Somatic ]
ABB Robotics and the Texas Children’s Hospital have developed a groundbreaking lab automation solution using ABB’s YuMi® cobot to transfer fruit flies (Drosophila melanogaster) used in the study for developing new drugs for neurological conditions such as Alzheimer’s, Huntington’s and Parkinson’s.
[ ABB ]
Extend Robotics are building embodied AI enabling highly flexible automation for real-world physical tasks. The system features intuitive immersive interface enabling tele-operation, supervision and training AI models.
[ Extend Robotics ]
The recorded livestream of RSS 2024 is now online, in case you missed anything.
[ RSS 2024 ]
-
IEEE and Keysight Team Up to Teach Kids About Electronics
by Robert Schneider on 15. August 2024. at 18:00
IEEE TryEngineering has partnered with Keysight Technologies to develop lesson plans focused on electronics and power simulation. Keysight provides hardware, software, and services to a wide variety of industries, particularly in the area of electronic measurement.
IEEE TryEngineering, an IEEE Educational Activities program, empowers educators to foster the next generation of technology innovators through free, online access to culturally relevant, developmentally appropriate, and educationally sound instructional resources for teachers and community volunteers.
The lesson plans cover a variety of STEM topics, experience levels, and age ranges. Educators should be able to find an applicable topic for their students, regardless of their grade level or interests.
Lesson plans on circuits
There are already a number of lesson plans available through the Keysight partnership that introduce students to electrical concepts, with more being developed. The most popular one thus far is Series and Parallel Circuits, which has been viewed more than 100 times each month. Teams of pupils predict the difference between a parallel and serial circuit design by building examples using wires, light bulbs, and batteries.
“TryEngineering is proud to be Keysight’s partner in attaining the ambitious goal of bringing engineering lessons to 1 million students in 2024.” —Debra Gulick
The newest of the Keysight-sponsored lesson plans, Light Up Name Badge, teaches the basics of circuitry, such as the components of a circuit, series and parallel circuits, and electronic component symbols. Students can apply their newfound knowledge in a design challenge wherein they create a light-up badge with their name.
Developing a workforce through STEM outreach
“Keysight’s commitment to workforce development through preuniversity STEM outreach makes it an ideal partner for IEEE TryEngineering,” says Debra Gulick, director of student and academic education programs for IEEE Educational Activities.
In addition, Keysight’s corporate social responsibility vision to build a better planet by accelerating innovation to connect and secure the world while employing a global business framework of ethical, environmentally sustainable, and socially responsible operations makes it a suitable IEEE partner.
“TryEngineering is proud to be Keysight’s partner in attaining the ambitious goal of bringing engineering lessons to 1 million students in 2024,” Gulick says.
The IEEE STEM Summit, a three-day virtual event in October for IEEE volunteers and educators, is expected to include a session highlighting Keysight’s lesson plans.
Educators and volunteers engaged in outreach activities with students can learn more on the Keysight TryEngineering partnership page.
The arrangement with Keysight was made possible with support from the IEEE Foundation. -
Optical Metasurfaces Shine a Light on Li-Fi, Lidar
by Margo Anderson on 15. August 2024. at 14:00
A new, tunable smart surface can transform a single pulse of light into multiple beams, each aimed in different directions. The proof-of-principle development opens the door to a range of innovations in communications, imaging, sensing, and medicine.
The research comes out of the Caltech lab of Harry Atwater, a professor of applied physics and materials science, and is possible due to a type of nano-engineered material called a metasurface. “These are artificially designed surfaces which basically consist of nanostructured patterns,” says Prachi Thureja, a graduate student in Atwater’s group. “So it’s an array of nanostructures, and each nanostructure essentially allows us to locally control the properties of light.”
The surface can be reconfigured up to millions of times per second to change how it is locally controlling light. That’s rapid enough to manipulate and redirect light for applications in optical data transmission such as optical space communications and Li-Fi, as well as lidar.
“[The metasurface] brings unprecedented freedom in controlling light,” says Alex M.H. Wong, an associate professor of electrical engineering at the City University of Hong Kong. “The ability to do this means one can migrate existing wireless technologies into the optical regime. Li-Fi and LIDAR serve as prime examples.”
Metasurfaces remove the need for lenses and mirrors
Manipulating and redirecting beams of light typically involves a range of conventional lenses and mirrors. These lenses and mirrors might be microscopic in size, but they’re still using optical properties of materials like Snell’s Law, which describes the progress of a wavefront through different materials and how that wavefront is redirected—or refracted—according to the properties of the material itself.
By contrast, the new work offers the prospect of electrically manipulating a material’s optical properties via a semiconducting material. Combined with nano-scaled mirror elements, the flat, microscopic devices can be made to behave like a lens, without requiring lengths of curved or bent glass. And the new metasurface’s optical properties can be switched millions of times per second using electrical signals.
“The difference with our device is by applying different voltages across the device, we can change the profile of light coming off of the mirror, even though physically it’s not moving,” says paper co-author Jared Sisler—also a graduate student in Atwater’s group. “And then we can steer the light like it’s an electrically reprogrammable mirror.”
The device itself, a chip that measures 120 micrometers on each side, achieves its light-manipulating capabilities with an embedded surface of tiny gold antennas in a semiconductor layer of indium tin oxide. Manipulating the voltages across the semiconductor alters the material’s capacity to bend light—also known as its index of refraction. Between the reflection of the gold mirror elements and the tunable refractive capacity of the semiconductor, a lot of rapidly-tunable light manipulation becomes possible.
“I think the whole idea of using a solid-state metasurface or optical device to steer light in space and also use that for encoding information—I mean, there’s nothing like that that exists right now,” Sisler says. “So I mean, technically, you can send more information if you can achieve higher modulation rates. But since it’s kind of a new domain, the performance of our device is more just to show the principle.”
Metasurfaces open up plenty of new possibilities
The principle, says Wong, suggests a wide array of future technologies on the back of what he says are likely near-term metasurface developments and discoveries.
“The metasurface [can] be flat, ultrathin, and lightweight while it attains the functions normally achieved by a series of carefully curved lenses,” Wong says. “Scientists are currently still unlocking the vast possibilities the metasurface has available to us.
“With improvements in nanofabrication, elements with small feature sizes much smaller than the wavelength are now reliably fabricable,” Wong continues. “Many functionalities of the metasurface are being routinely demonstrated, benefiting not just communication but also imaging, sensing, and medicine, among other fields... I know that in addition to interest from academia, various players from industry are also deeply interested and making sizable investments in pushing this technology toward commercialization.”
-
NIST Announces Post-Quantum Cryptography Standards
by Dina Genkina on 13. August 2024. at 10:01
Today, almost all data on the Internet, including bank transactions, medical records, and secure chats, is protected with an encryption scheme called RSA (named after its creators Rivest, Shamir, and Adleman). This scheme is based on a simple fact—it is virtually impossible to calculate the prime factors of a large number in a reasonable amount of time, even on the world’s most powerful supercomputer. Unfortunately, large quantum computers, if and when they are built, would find this task a breeze, thus undermining the security of the entire Internet.
Luckily, quantum computers are only better than classical ones at a select class of problems, and there are plenty of encryption schemes where quantum computers don’t offer any advantage. Today, the U.S. National Institute of Standards and Technology (NIST) announced the standardization of three post-quantum cryptography encryption schemes. With these standards in hand, NIST is encouraging computer system administrators to begin transitioning to post-quantum security as soon as possible.
“Now our task is to replace the protocol in every device, which is not an easy task.” —Lily Chen, NIST
These standards are likely to be a big element of the Internet’s future. NIST’s previous cryptography standards, developed in the 1970s, are used in almost all devices, including Internet routers, phones, and laptops, says Lily Chen, head of the cryptography group at NIST who lead the standardization process. But adoption will not happen overnight.
“Today, public key cryptography is used everywhere in every device,” Chen says. “Now our task is to replace the protocol in every device, which is not an easy task.”
Why we need post-quantum cryptography now
Most experts believe large-scale quantum computers won’t be built for at least another decade. So why is NIST worried about this now? There are two main reasons.
First, many devices that use RSA security, like cars and some IoT devices, are expected to remain in use for at least a decade. So they need to be equipped with quantum-safe cryptography before they are released into the field.
“For us, it’s not an option to just wait and see what happens. We want to be ready and implement solutions as soon as possible.” —Richard Marty, LGT Financial Services
Second, a nefarious individual could potentially download and store encrypted data today, and decrypt it once a large enough quantum computer comes online. This concept is called “harvest now, decrypt later“ and by its nature, it poses a threat to sensitive data now, even if that data can only be cracked in the future.
Security experts in various industries are starting to take the threat of quantum computers seriously, says Joost Renes, principal security architect and cryptographer at NXP Semiconductors. “Back in 2017, 2018, people would ask ‘What’s a quantum computer?’” Renes says. “Now, they’re asking ‘When will the PQC standards come out and which one should we implement?’”
Richard Marty, chief technology officer at LGT Financial Services, agrees. “For us, it’s not an option to just wait and see what happens. We want to be ready and implement solutions as soon as possible, to avoid harvest now and decrypt later.”
NIST’s competition for the best quantum-safe algorithm
NIST announced a public competition for the best PQC algorithm back in 2016. They received a whopping 82 submissions from teams in 25 different countries. Since then, NIST has gone through 4 elimination rounds, finally whittling the pool down to four algorithms in 2022.
This lengthy process was a community-wide effort, with NIST taking input from the cryptographic research community, industry, and government stakeholders. “Industry has provided very valuable feedback,” says NIST’s Chen.
These four winning algorithms had intense-sounding names: CRYSTALS-Kyber, CRYSTALS-Dilithium, Sphincs+, and FALCON. Sadly, the names did not survive standardization: The algorithms are now known as Federal Information Processing Standard (FIPS) 203 through 206. FIPS 203, 204, and 205 are the focus of today’s announcement from NIST. FIPS 206, the algorithm previously known as FALCON, is expected to be standardized in late 2024.
The algorithms fall into two categories: general encryption, used to protect information transferred via a public network, and digital signature, used to authenticate individuals. Digital signatures are essential for preventing malware attacks, says Chen.
Every cryptography protocol is based on a math problem that’s hard to solve but easy to check once you have the correct answer. For RSA, it’s factoring large numbers into two primes—it’s hard to figure out what those two primes are (for a classical computer), but once you have one it’s straightforward to divide and get the other.
“We have a few instances of [PQC], but for a full transition, I couldn’t give you a number, but there’s a lot to do.” —Richard Marty, LGT Financial Services
Two out of the three schemes already standardized by NIST, FIPS 203 and FIPS 204 (as well as the upcoming FIPS 206), are based on another hard problem, called lattice cryptography. Lattice cryptography rests on the tricky problem of finding the lowest common multiple among a set of numbers. Usually, this is implemented in many dimensions, or on a lattice, where the least common multiple is a vector.
The third standardized scheme, FIPS 205, is based on hash functions—in other words, converting a message to an encrypted string that’s difficult to reverse
The standards include the encryption algorithms’ computer code, instructions for how to implement it, and intended uses. There are three levels of security for each protocol, designed to future-proof the standards in case some weaknesses or vulnerabilities are found in the algorithms.
Lattice cryptography survives alarms over vulnerabilities
Earlier this year, a pre-print published to the arXiv alarmed the PQC community. The paper, authored by Yilei Chen of Tsinghua University in Beijing, claimed to show that lattice-based cryptography, the basis of two out of the three NIST protocols, was not, in fact, immune to quantum attacks. On further inspection, Yilei Chen’s argument turned out to have a flaw—and lattice cryptography is still believed to be secure against quantum attacks.
On the one hand, this incident highlights the central problem at the heart of all cryptography schemes: There is no proof that any of the math problems the schemes are based on are actually “hard.” The only proof, even for the standard RSA algorithms, is that people have been trying to break the encryption for a long time, and have all failed. Since post-quantum cryptography standards, including lattice cryptography, are newer, there is less certainty that no one will find a way to break them.
That said, the failure of this latest attempt only builds on the algorithm’s credibility. The flaw in the paper’s argument was discovered within a week, signaling that there is an active community of experts working on this problem. “The result of that paper is not valid, that means the pedigree of the lattice-based cryptography is still secure,” says NIST’s Lily Chen (no relation to Tsinghua University’s Yilei Chen). “People have tried hard to break this algorithm. A lot of people are trying, they try very hard, and this actually gives us confidence.”
NIST’s announcement is exciting, but the work of transitioning all devices to the new standards has only just begun. It is going to take time, and money, to fully protect the world from the threat of future quantum computers.
“We’ve spent 18 months on the transition and spent about half a million dollars on it,” says Marty of LGT Financial Services. “We have a few instances of [PQC], but for a full transition, I couldn’t give you a number, but there’s a lot to do.”
-
Level Up Your Leadership Skills with IEEE Courses
by Nicholas Spada on 12. August 2024. at 19:00
Author and leadership expert John C. Maxwell famously said, “The single biggest way to impact an organization is to focus on leadership development. There is almost no limit to the potential of an organization that recruits good people, raises them up as leaders, and continually develops them.”
Experts confirm that there are clear benefits to fostering leadership by encouraging employees’ professional growth and nurturing and developing company leaders. A culture of leadership development and innovation boosts employee engagement by 20 percent to 25 percent, according to an analysis in the Journal of Applied Psychology. Companies are 22 percent more profitable, on average, when they engage their employees by building a culture of leadership, innovation, and recognition, according to Zippia research.
Developing professionals into strong leaders can have a lasting impact on a company, and the IEEE Professional Development Suite can help make it possible. The training programs in the suite help aspiring technology leaders who want to develop their essential business and management skills. Programs include IEEE Leading Technical Teams, the IEEE | Rutgers Online Mini-MBA for Engineers and Technical Professionals, and the Intensive Wireless Communications and Advanced Topics in Wireless courses offered by the IEEE Communications Society. IEEE also offers topical courses through its eLearning Library.
Tips for leading teams
IEEE Leading Technical Teams is a live, six-hour course offered both in person and virtually. Addressing challenges that come with leading groups, it is designed for team leaders, managers, and directors of engineering and technical teams.
“Participating benefited me and my employer by enhancing my leadership skills in inspiring others to achieve the goals of our organization,” says Stephen Wilkowski, a system test engineer at CACI International in Reston, Va., who completed the training. “I found the leadership practices assessment to be very valuable, as I appreciated the anonymous feedback received from those who I work with. I would recommend the training to anyone desiring to improve their leadership skills.”
Attendees participate in the 360° Leadership Practices Inventory, a tool that solicits confidential feedback on the participant’s strengths and opportunities for improvement from their team members and managers. The program encompasses instructor-led exercises and case studies demonstrating the application of best practices to workplace challenges.
Participants learn the “five practices of exemplary leadership” and receive valuable peer coaching.
To learn more about in-person and virtual options for individuals and companies, complete this form.
A mini-MBA for technologists
The 12-week IEEE | Rutgers Online Mini-MBA for Engineers and Technical Professionals program covers business strategy, new product development management, financial analysis, sales and marketing, and leadership. It includes a combination of expert instruction, peer interaction, self-paced video lessons, interactive assessments, live office hours, and hands-on capstone project experience. The program offers flexible learning opportunities for individual learners as well as customized company cohort options.
Developing professionals into strong leaders can have a lasting impact on a company, and the IEEE Professional Development Suite can help make that possible.
“The mini-MBA was a great opportunity to explore other areas of business that I don’t typically encounter,” says graduate Jonathan Bentz, a senior manager at Nvidia. “I have a customer-facing technical role, and the mini-MBA allowed me to get a taste of the full realm of business leadership.”
For more information, see IEEE | Rutgers Online Mini-MBA for Engineers and Technical Professionals.
Training on wireless communications
The Intensive Wireless Communications and the Advanced Topics in Wireless course series are exclusively presented by the IEEE Communications Society.
The Intensive Wireless interactive live course provides training necessary to stay on top of key developments in the dynamic, rapidly evolving communications industry. Designed for those with an engineering background who want to enhance their knowledge of wireless communication technologies, the series is an ideal way to train individual employees or your entire team at once.
The Advanced Topics in Wireless series is for engineers and technical professionals with a working knowledge of wireless who are looking to enhance their skill set. The series dives into recent advancements, applications, and use cases in emerging connectivity.
Participants in the live, online course series develop a comprehensive view of 5G/NR technology, as well as an understanding of the implementation of all the ITU-specified use case categories such as enhanced mobile broadband, mIoT, and ultra-reliable low-latency communication. The series also provides a robust foundation on the network architecture and the evolution of technology, which enables fully open radio access networks.
Learn more about the Advanced Topics in Wireless Course Series by completing this form.
Topics in the eLearning Library
Tailored for professionals, faculty, and students, the IEEE eLearning Library taps into a wealth of expertise from the organization’s global network of more than 450,000 industry and academia members. Courses cover a wide variety of disciplines including artificial intelligence, blockchain technology, cyber and data security, power and energy, telecommunications, and IEEE standards.
You can help foster growth and leadership skills for your organization by offering employees access to hundreds of courses. Start exploring the library by filling out this form.
Completion of course programs offers learners the ability to earn IEEE certificates bearing professional development hours, continuing education units, and digital badges.
-
Amazon Vies for Nuclear-Powered Data Center
by Andrew Moseman on 12. August 2024. at 18:36
When Amazon Web Services paid US $650 million in March for another data center to add to its armada, the tech giant thought it was buying a steady supply of nuclear energy to power it, too. The Susquehanna Steam Electric Station outside of Berick, Pennsylvania, which generates 2.5 gigawatts of nuclear power, sits adjacent to the humming data center and had been directly powering it since the center opened in 2023.
After striking the deal, Amazon wanted to change the terms of its original agreement to buy 180 megawatts of additional power directly from the nuclear plant. Susquehanna agreed to sell it. But third parties weren’t happy about that, and their deal has become bogged down in a regulatory battle that will likely set a precedent for data centers, cryptocurrency mining operations, and other computing facilities with voracious appetites for clean electricity.
Putting a data center right next to a power plant so that it can draw electricity from it directly, rather than from the grid, is becoming more common as data centers seek out cheap, steady, carbon-free power. Proposals for co-locating data centers next to nuclear power have popped up in New Jersey, Texas, Ohio, and elsewhere. Sweden is considering using small modular reactors to power future data centers.
However, co-location raises questions about equity and energy security, because directly-connected data centers can avoid paying fees that would otherwise help maintain grids. They also hog hundreds of megawatts that could be going elsewhere.
“They’re effectively going behind the meter and taking that capacity off of the grid that would otherwise serve all customers,” says Tony Clark, a senior advisor at the law firm Wilkinson Barker Knauer and a former commissioner at the Federal Energy Regulatory Commission (FERC), who has testified to a U.S. House subcommittee on the subject.
Amazon’s nuclear power deal meets hurdles
The dust-up over the Amazon-Susquehanna agreement started in June, after Amazon subsidiary Amazon Web Services filed a notice to change its interconnection service agreement (ISA) in order to buy more nuclear power from Susquehanna’s parent company, Talen Energy. Amazon wanted to increase the amount of behind-the-meter power it buys from the plant from 300 MW to 480 MW. Shortly after it requested the change, utility giants Exelon and American Electric Power (AEP), filed a protest against the agreement and asked FERC to hold a hearing on the matter.
Their complaint: the deal between Amazon and the nuclear plant would hurt a third party, namely all the customers who buy power from AEP or Exelon utilities. The protest document argues that the arrangement would shift up to $140 million in extra costs onto the people of Pennsylvania, New Jersey, and other states served by PJM, a regional transmission organization that oversees the grid in those areas. “Multiplied by the many similar projects on the drawing board, it is apparent that this unsupported filing has huge financial consequences that should not be imposed on ratepayers without sufficient process to determine and evaluate what is really going on,” their complaint says.
Susquehanna dismissed the argument, effectively saying that its deal with Amazon is none of AEP and Exelon’s business. “It is an unlawful attempt to hijack this limited [ISA] amendment proceeding that they have no stake in and turn it into an ad hoc national referendum on the future of data center load,” Susquehanna’s statement said. (AEP, Exelon, Talen/Susquehanna, and Amazon all declined to comment for this story.)
More disputes like this will likely follow as more data centers co-locate with clean energy. Kevin Schneider, a power system expert at Pacific Northwest National Laboratory and research professor at Washington State University, says it’s only natural that data center operators want the constant, consistent nature of nuclear power. “If you look at the base load nature of nuclear, you basically run it up to a power level and leave it there. It can be well aligned with a server farm.”
Data center operators are also exploring energy options from solar and wind, but these energy sources would have a difficult time matching the constancy of nuclear, even with grid storage to help even out their supply. So giant tech firms look to nuclear to keep their servers running without burning fossil fuels, and use that to trumpet their carbon-free achievements, as Amazon did when it bought the data center in Pennsylvania. “Whether you’re talking about Google or Apple or Microsoft or any of those companies, they tend to have corporate sustainability goals. Being served by a nuclear unit looks great on their corporate carbon balance sheet,” Clark says.
Costs of data centers seeking nuclear energy
Yet such arrangements could have major consequences for other energy customers, Clark argues. For one, directing all the energy from a nuclear plant to a data center is, fundamentally, no different than retiring that plant and taking it offline. “It’s just a huge chunk of capacity leaving the system,” he says, resulting in higher prices and less energy supply for everyone else.
Another issue is the “behind-the-meter” aspect of these kinds of deals. A data center could just connect to the grid and draw from the same supply as everyone else, Clark says. But by connecting directly to the power plant, the center’s owner avoids paying the administrative fees that are used to maintain the grid and grow its infrastructure. Those costs could then get passed on to businesses and residents who have to buy power from the grid. “There’s just a whole list of charges that get assessed through the network service that if you don’t connect through the network, you don’t have to pay,” Clark says. “And those charges are the part of the bill that will go up” for everyone else.
Even the “carbon-free” public relations talking points that come with co-location may be suspect in some cases. In Washington State, where Schneider works, new data centers are being planted next to the region’s abundant hydropower stations, and they’re using so much of that energy that parts of the state are considering adding more fossil fuel capacity to make ends meet. This results in a “zero-emissions shell game,” Clark wrote in a white paper on the subject.
These early cases are likely only the beginning. A report posted in May from the Electric Power Research Institute predicts energy demand from data centers will double by 2030, a leap driven by the fact that AI queries need ten times more energy than traditional internet searches. The International Energy Agency puts the timeline for doubling sooner–in 2026. Data centers, AI, and the cryptocurrency sector consumed an estimated 460 terawatt-hours (TWh) in 2022, and could reach more than 1000 TWh in 2026, the agency predicts.
Data centers face energy supply challenges
New data centers can be built in a matter of months, but it takes years to build utility-scale power projects, says Poorvi Patel, manager of strategic insights at Electric Power Research Institute and contributor to the report. The potential for unsustainable growth in electricity needs has put grid operators on alert, and in some cases has sent them sounding the alarm. Eirgrid, a state-owned transmission operator in Ireland, last week warned of a “mass exodus” of data centers in Ireland if it can’t connect new sources of energy.
There’s only so much existing nuclear power to go around, and enormous logistical and regulatory roadblocks to building more. So data center operators and tech giants are looking for creative solutions. Some are considering small modular reactors (SMRs)–which are advanced nuclear reactors with much smaller operating capacities than conventional reactors. Nano Nuclear Energy, which is developing microreactors–a particularly small type of SMR–last month announced an agreement with Blockfusion to explore the possibility of powering a currently defunct cryptomining facility in Niagara Falls, New York.
“To me, it does seem like a space where, if big tech has a voracious electric power needs and they really want that 24/7, carbon-free power, nuclear does seem to be the answer,” Clark says. “They also have the balance sheets to be able to do some of the risk mitigation that might make it attractive to get an SMR up and running.”
-
Photonic Chip Cuts Cost of Hunting Exoplanets
by Rachel Berkowitz on 12. August 2024. at 13:01
At 6.5 meters in diameter, the James Webb Space Telescope’s primary mirror captures more light than any telescope that’s ever been launched from Earth. But not every astronomer has US $10 billion to spend on a space telescope. So to help bring the cost of space-based astronomy down, researchers at the National Research Council of Canada in Ottawa are working on a way to process starlight on a tiny optical chip. Ross Cheriton, a photonics researcher there, and his students built and tested a CubeSat prototype with a new kind of photonic chip. The goal is to lower the barrier to entry for astronomical science using swarms of lower-cost spacecraft.
“We hope to enable smaller space telescopes to do big science using highly compact instrument-on-chips,” Cheriton says, who is also affiliated with the Quantum and Nanotechnology Research Centre in Ottawa.
Photonics integrated circuits (PICs) use light instead of electricity to process information, and they’re in wide use slinging trillions and trillions of bits around data centers. But only recently have astronomers begun to examine how to use them to push the boundaries of what can be learned about the universe.
Ground-based telescopes are plagued by Earth’s atmosphere, where turbulence blurs incoming light, making it difficult to focus it onto a camera chip. In outer space, telescopes can peer at extremely faint objects in non-visible wavelengths without correcting for the impact of turbulence. That’s where Cheriton aims to boldly go with a PIC filter that detects very subtle gas signatures during an exoplanet “eclipse” called a transit.
The main motivation for putting photonic chips in space is to reduce the size, weight, and cost of components, because it can be produced en masse in a semiconductor foundry. “The dream is a purely fiber and chip-based instrument with no other optics,” says Cheriton. Replacing filters, lenses, and mirrors with a chip also improves stability and scalability compared to ordinary optical parts.
CubeSats—inexpensive, small, and standardized satellites—have proved to be a cost-effective way of deploying small instrument payloads. “The compact nature of PICs is a perfect match for CubeSats to study bright exoplanet systems James Webb doesn’t have time to stare at,” says Cheriton.
For a total mission cost of less than $1 million—compared to the Webb’s $10 billion—an eventual CubeSat mission could stare at a star for days to weeks while it waits for a planet to cross the field of view. Then, it would look for slight changes in the star’s spectrum that are associated with how the planet’s atmosphere absorbs light—telltale evidence of gasses of a biological origin.
Smaller spectroscopy
As a proof-of-concept, Cheriton guided a team of undergraduate students who spent eight months designing and integrating a PIC into a custom 3U CubeSat (10 centimeter x 10 cm x 30 cm) platform. Their silicon nitride photonic circuit sensor proved itself capable of detecting the absorption signatures of CO2 in incoming light.
In their design, light entering the CubeSat’s collimating lens gets focused into a fiber and then pushed to the photonic chip. It enters an etched set of waveguides that includes a ring resonator. Here, light having a specific set of wavelengths builds in intensity over multiple trips around the ring, and is then output to a detector. Because only a select few wavelengths constructively interfere—those chosen to match a gas’s absorption spectrum—the ring serves as a comb-like filter. After the light goes through the ring resonator, the signal from the waveguide gets passed to an output fiber and onto a camera connected to a Raspberry Pi computer for processing. A single pixel’s intensity therefore serves as a reading for a gas’s presence.
Light travels through a waveguide on a photonic integrated circuit.Teseract
Because it’s built on a chip, the sensor could be multiplexed for observing several objects or sense different gasses simultaneously. Additionally, all the light falling on a single pixel means that the signal is more sensitive than a traditional spectrometer, says Cheriton. Moreover, instead of hunting for peaks in a full spectrum, the technology looks for how well the absorption spectrum matches that of a specific gas, a more efficient process. “If something is in space, you don’t want to send gigabytes of data home if you don’t have to,” he says.
Space travel is still a long way off for the astrophotonic CubeSat. The current design does not use space-qualified components. But Cheriton’s students tested it in the lab for red light (635 nm) and CO2 in a gas cell. They used a “ground station” computer to transmit all commands and receive all results—and to monitor the photovoltaics and collect data from the flight control sensors onboard their CubeSat.
Next, the team plans to test whether their sensor can detect oxygen with the silicon nitride chip, a material that was chosen for its transparency to the gas’s 760 nm wavelength. Success would leave them well positioned to meet what Cheriton calls the next huge milestone for astronomers: looking for an earth-like planet with oxygen.
The work was presented at the Optica (formerly Optical Society of America) Advanced Photonics conference in July.
-
Hybrid Bonding Plays Starring Role in 3D Chips
by Samuel K. Moore on 11. August 2024. at 13:00
Chipmakers continue to claw for every spare nanometer to continue scaling down circuits, but a technology involving things that are much bigger—hundreds or thousands of nanometers across—could be just as significant over the next five years.
Called hybrid bonding, that technology stacks two or more chips atop one another in the same package. That allows chipmakers to increase the number of transistors in their processors and memories despite a general slowdown in the shrinking of transistors, which once drove Moore’s Law. At the IEEE Electronic Components and Technology Conference (ECTC) this past May in Denver, research groups from around the world unveiled a variety of hard-fought improvements to the technology, with a few showing results that could lead to a record density of connections between 3D stacked chips: some 7 million links per square millimeter of silicon.
All those connections are needed because of the new nature of progress in semiconductors, Intel’s Yi Shi told engineers at ECTC. Moore’s Law is now governed by a concept called system technology co-optimization, or STCO, whereby a chip’s functions, such as cache memory, input/output, and logic, are fabricated separately using the best manufacturing technology for each. Hybrid bonding and other advanced packaging tech can then be used to assemble these subsystems so that they work every bit as well as a single piece of silicon. But that can happen only when there’s a high density of connections that can shuttle bits between the separate pieces of silicon with little delay or energy consumption.
Out of all the advanced-packaging technologies, hybrid bonding provides the highest density of vertical connections. Consequently, it is the fastest growing segment of the advanced-packaging industry, says Gabriella Pereira, technology and market analyst at Yole Group. The overall market is set to more than triple to US $38 billion by 2029, according to Yole, which projects that hybrid bonding will make up about half the market by then, although today it’s just a small portion.
In hybrid bonding, copper pads are built on the top face of each chip. The copper is surrounded by insulation, usually silicon oxide, and the pads themselves are slightly recessed from the surface of the insulation. After the oxide is chemically modified, the two chips are then pressed together face-to-face, so that the recessed pads on each align. This sandwich is then slowly heated, causing the copper to expand across the gap and fuse, connecting the two chips.
Making Hybrid Bonding Better
- Hybrid bonding starts with two wafers or a chip and a wafer facing each other. The mating surfaces are covered in oxide insulation and slightly recessed copper pads connected to the chips’ interconnect layers.
- The wafers are pressed together to form an initial bond between the oxides.
- The stacked wafers are then heated slowly, strongly linking the oxides and expanding the copper to form an electrical connection.
- To form more secure bonds, engineers are flattening the last few nanometers of oxide. Even slight bulges or warping can break dense connections.
- The copper must be recessed from the surface of the oxide just the right amount. Too much and it will fail to form a connection. Too little and it will push the wafers apart. Researchers are working on ways to control the level of copper down to single atomic layers.
- The initial links between the wafers are weak hydrogen bonds. After annealing, the links are strong covalent bonds [below]. Researchers expect that using different types of surfaces, such as silicon carbonitride, which has more locations to form chemical bonds, will lead to stronger links between the wafers.
- The final step in hybrid bonding can take hours and require high temperatures. Researchers hope to lower the temperature and shorten the process time.
- Although the copper from both wafers presses together to form an electrical connection, the metal’s grain boundaries generally do not cross from one side to the other. Researchers are trying to cause large single grains of copper to form across the boundary to improve conductance and stability.
Hybrid bonding can either attach individual chips of one size to a wafer full of chips of a larger size or bond two full wafers of chips of the same size. Thanks in part to its use in camera chips, the latter process is more mature than the former, Pereira says. For example, engineers at the European microelectronics-research institute Imec have created some of the most dense wafer-on-wafer bonds ever, with a bond-to-bond distance (or pitch) of just 400 nanometers. But Imec managed only a 2-micrometer pitch for chip-on-wafer bonding.
The latter is a huge improvement over the advanced 3D chips in production today, which have connections about 9 μm apart. And it’s an even bigger leap over the predecessor technology: “microbumps” of solder, which have pitches in the tens of micrometers.
“With the equipment available, it’s easier to align wafer to wafer than chip to wafer. Most processes for microelectronics are made for [full] wafers,” says Jean-Charles Souriau, scientific leader in integration and packaging at the French research organization CEA Leti. But it’s chip-on-wafer (or die-to-wafer) that’s making a splash in high-end processors such as those from AMD, where the technique is used to assemble compute cores and cache memory in its advanced CPUs and AI accelerators.
In pushing for tighter and tighter pitches for both scenarios, researchers are focused on making surfaces flatter, getting bound wafers to stick together better, and cutting the time and complexity of the whole process. Getting it right could revolutionize how chips are designed.
WoW, Those Are Some Tight Pitches
The recent wafer-on-wafer (WoW) research that achieved the tightest pitches—from 360 nm to 500 nm—involved a lot of effort on one thing: flatness. To bond two wafers together with 100-nm-level accuracy, the whole wafer has to be nearly perfectly flat. If it’s bowed or warped to the slightest degree, whole sections won’t connect.
Flattening wafers is the job of a process called chemical mechanical planarization, or CMP. It’s essential to chipmaking generally, especially for producing the layers of interconnects above the transistors.
“CMP is a key parameter we have to control for hybrid bonding,” says Souriau. The results presented at ECTC show CMP being taken to another level, not just flattening across the wafer but reducing mere nanometers of roundness on the insulation between the copper pads to ensure better connections.
“It’s difficult to say what the limit will be. Things are moving very fast.” —Jean-Charles Souriau, CEA Leti
Other researchers focused on ensuring those flattened parts stick together strongly enough. They did so by experimenting with different surface materials such as silicon carbonitride instead of silicon oxide and by using different schemes to chemically activate the surface. Initially, when wafers or dies are pressed together, they are held in place with relatively weak hydrogen bonds, and the concern is whether everything will stay in place during further processing steps. After attachment, wafers and chips are then heated slowly, in a process called annealing, to form stronger chemical bonds. Just how strong these bonds are—and even how to figure that out—was the subject of much of the research presented at ECTC.
Part of that final bond strength comes from the copper connections. The annealing step expands the copper across the gap to form a conductive bridge. Controlling the size of that gap is key, explains Samsung’s Seung Ho Hahn. Too little expansion, and the copper won’t fuse. Too much, and the wafers will be pushed apart. It’s a matter of nanometers, and Hahn reported research on a new chemical process that he hopes to use to get it just right by etching away the copper a single atomic layer at a time.
The quality of the connection counts, too. The metals in chip interconnects are not a single crystal; instead they’re made up of many grains, crystals oriented in different directions. Even after the copper expands, the metal’s grain boundaries often don’t cross from one side to another. Such a crossing should reduce a connection’s electrical resistance and boost its reliability. Researchers at Tohoku University in Japan reported a new metallurgical scheme that could finally generate large, single grains of copper that cross the boundary. “This is a drastic change,” says Takafumi Fukushima, an associate professor at Tohoku. “We are now analyzing what underlies it.”
Other experiments discussed at ECTC focused on streamlining the bonding process. Several sought to reduce the annealing temperature needed to form bonds—typically around 300 °C—as to minimize any risk of damage to the chips from the prolonged heating. Researchers from Applied Materials presented progress on a method to radically reduce the time needed for annealing—from hours to just 5 minutes.
CoWs That Are Outstanding in the Field
Imec used plasma etching to dice up chips and give them chamfered corners. The technique relieves mechanical stress that could interfere with bonding.Imec
Chip-on-wafer (CoW) hybrid bonding is more useful to makers of advanced CPUs and GPUs at the moment: It allows chipmakers to stack chiplets of different sizes and to test each chip before it’s bound to another, ensuring that they aren’t dooming an expensive CPU with a single flawed part.
But CoW comes with all of the difficulties of WoW and fewer of the options to alleviate them. For example, CMP is designed to flatten wafers, not individual dies. Once dies have been cut from their source wafer and tested, there’s less that can be done to improve their readiness for bonding.
Nevertheless, researchers at Intel reported CoW hybrid bonds with a 3-μm pitch, and, as mentioned, a team at Imec managed 2 μm, largely by making the transferred dies very flat while they were still attached to the wafer and keeping them extra clean throughout the process. Both groups used plasma etching to dice up the dies instead of the usual method, which uses a specialized blade. Unlike a blade, plasma etching doesn’t lead to chipping at the edges, which creates debris that could interfere with connections. It also allowed the Imec group to shape the die, making chamfered corners that relieve mechanical stress that could break connections.
CoW hybrid bonding is going to be critical to the future of high-bandwidth memory (HBM), according to several researchers at ECTC. HBM is a stack of DRAM dies—currently 8 to 12 dies high—atop a control-logic chip. Often placed within the same package as high-end GPUs, HBM is crucial to handling the tsunami of data needed to run large language models like ChatGPT. Today, HBM dies are stacked using microbump technology, so there are tiny balls of solder surrounded by an organic filler between each layer.
But with AI pushing memory demand even higher, DRAM makers want to stack 20 layers or more in HBM chips. The volume that microbumps take up means that these stacks will soon be too tall to fit properly in the package with GPUs. Hybrid bonding would shrink the height of HBMs and also make it easier to remove excess heat from the package, because there would be less thermal resistance between its layers.
“I think it’s possible to make a more-than-20-layer stack using this technology.” —Hyeonmin Lee, Samsung
At ECTC, Samsung engineers showed that hybrid bonding could yield a 16-layer HBM stack. “I think it’s possible to make a more-than-20-layer stack using this technology,” says Hyeonmin Lee, a senior engineer at Samsung. Other new CoW technology could also help bring hybrid bonding to high-bandwidth memory. Researchers at CEA Leti are exploring what’s known as self-alignment technology, says Souriau. That would help ensure good CoW connections using just chemical processes. Some parts of each surface would be made hydrophobic and some hydrophilic, resulting in surfaces that would slide into place automatically.
At ECTC, researchers from Tohoku University and Yamaha Robotics reported work on a similar scheme, using the surface tension of water to align 5-μm pads on experimental DRAM chips with better than 50-nm accuracy.
The Bounds of Hybrid Bonding
Researchers will almost certainly keep reducing the pitch of hybrid-bonding connections. A 200-nm WoW pitch is not just possible but desirable, Han-Jong Chia, a project manager for pathfinding systems at Taiwan Semiconductor Manufacturing Co. , told engineers at ECTC. Within two years, TSMC plans to introduce a technology called backside power delivery. (Intel plans the same for the end of this year.) That’s a technology that puts the chip’s chunky power-delivery interconnects below the surface of the silicon instead of above it. With those power conduits out of the way, the uppermost levels can connect better to smaller hybrid-bonding bond pads, TSMC researchers calculate. Backside power delivery with 200-nm bond pads would cut down the capacitance of 3D connections so much that a measure of energy efficiency and signal speed would be as much as eight times better than what can be achieved with 400-nm bond pads.
Chip-on-wafer hybrid bonding is more useful than wafer-on-wafer bonding, in that it can place dies of one size onto a wafer of larger dies. However, the density of connections that can be achieved is lower than for wafer-on-wafer bonding.Imec
At some point in the future, if bond pitches narrow even further, Chia suggests, it might become practical to “fold” blocks of circuitry so they are built across two wafers. That way some of what are now long connections within the block might be able to take a vertical shortcut, potentially speeding computations and lowering power consumption.
And hybrid bonding may not be limited to silicon. “Today there is a lot of development in silicon-to-silicon wafers, but we are also looking to do hybrid bonding between gallium nitride and silicon wafers and glass wafers…everything on everything,” says CEA Leti’s Souriau. His organization even presented research on hybrid bonding for quantum-computing chips, which involves aligning and bonding superconducting niobium instead of copper.
“It’s difficult to say what the limit will be,” Souriau says. “Things are moving very fast.”
This article was updated on 11 August 2024.
This article appears in the September 2024 print issue as “The Copper Connection.”
-
Trailblazing Tech Leader Helps Shape U.S. AI Strategy
by Joanna Goodrich on 10. August 2024. at 13:00
In the two years since Arati Prabhakar was appointed director of the White House Office of Science and Technology Policy, she has set the United States on a course toward regulating artificial intelligence. The IEEE Fellow advised the U.S. President Joe Biden in writing the executive order he issued to accomplish the goal just six months after she began her new role in 2022.
Prabhakar is the first woman and the first person of color to serve as OSTP director, and she has broken through the glass ceiling at other agencies as well. She was the first woman to lead the National Institute of Standards and Technology (NIST) and the Defense Advanced Research Projects Agency.
Arati Prabhakar
Employer
U.S. government
Title
Director of the White House Office of Science and Technology Policy
Member grade
Fellow
Alma maters
Texas Tech University; Caltech
Working in the public sector wasn’t initially on her radar. Not until she became a DARPA program manager in 1986, she says, did she really understand what she could accomplish as a government official.
“What I have come to love about [public service] is the opportunity to shape policies at a scale that is really unparalleled,” she says.
Prabhakar’s passion for tackling societal challenges by developing technology also led her to take leadership positions at companies including Raychem (now part of TE Connectivity), Interval Research Corp., and U.S. Venture Partners. In 2019 she helped found Actuate, a nonprofit in Palo Alto, Calif., that seeks to create technology to help address climate change, data privacy, health care access, and other pressing issues.
“I really treasure having seen science, technology, and innovation from all different perspectives,” she says. “But the part I have loved most is public service because of the impact and reach that it can have.”
Discovering her passion for electrical engineering
Prabhakar, who was born in India and raised in Texas, says she decided to pursue a STEM career because when she was growing up, her classmates said women weren’t supposed to work in science, technology, engineering or mathematics.
“Them saying that just made me want to pursue it more,” she says. Her parents, who had wanted her to become a doctor, supported her pursuit of engineering, she adds.
After earning a bachelor’s degree in electrical engineering in 1979 from Texas Tech University, in Lubbock, she moved to California to continue her education at Caltech. She graduated with a master’s degree in EE in 1980, then earned a doctorate in applied physics in 1984. Her doctoral thesis focused on understanding deep-level defects and impurities in semiconductors that affect device performance.
After acquiring her Ph.D., she says, she wanted to make a bigger impact with her research than academia would allow, so she applied for a policy fellowship from the American Association for the Advancement of Science to work at the congressional Office of Technology Assessment. The office examines issues involving new or expanding technologies, assesses their impact, and studies whether new policies are warranted.
“We have huge aspirations for the future—such as mitigating climate change—that science and technology have to be part of achieving.”
“I wanted to share my research in semiconductor manufacturing processes with others,” Prabhakar says. “That’s what felt exciting and valuable to me.”
She was accepted into the program and moved to Washington, D.C. During the yearlong fellowship, she conducted a study on microelectronics R&D for the research and technology subcommittee of the U.S. House of Representatives committee on science, space, and technology. The subcommittee oversees STEM-related matters including education, policy, and standards.
While there, she worked with people who were passionate about public service and government, but she didn’t feel the same, she says, until she joined DARPA. As program manager, Prabhakar established and led several projects including a microelectronics office that invests in developing new technologies in areas such as lithography, optoelectronics, infrared imaging, and neural networks.
In 1993 an opportunity arose that she couldn’t refuse, she says: President Bill Clinton nominated her to direct the National Institute of Standards and Technology. NIST develops technical guidelines and conducts research to create tools that improve citizens’ quality of life. At age 34, she became the first woman to lead the agency.
Believing in IEEE’s Mission
Like many IEEE members, Prabhakar says, she joined IEEE as a student member while attending Texas Tech University because the organization’s mission aligned with her belief that engineering is about creating value in the world.
She continues to renew her membership, she says, because IEEE emphasizes that technology should benefit humanity.
“It really comes back to this idea of the purpose of engineering and the role that it plays in the world,” she says.
After leading NIST through the first Clinton administration, she left for the private sector, including stints as CTO at appliance-component maker Raychem in Menlo Park, Calif., and president of private R&D lab Interval Research of Palo Alto, Calif. In all, she spent the next 14 years in the private sector, mostly as a partner at U.S. Venture Partners, in Menlo Park, where she invested in semiconductor and clean-tech startups.
In 2012 she returned to DARPA and became its first female director.
“When I received the call offering me the job, I stopped breathing,” Prabhakar says. “It was a once-in-a-lifetime opportunity to make a difference at an agency that I had loved earlier in my career. And it proved to be just as meaningful an experience as I had hoped.”
For the next five years she led the agency, focusing on developing better military systems and the next generation of artificial intelligence, as well as creating solutions in social science, synthetic biology, and neurotechnology.
Under her leadership, in 2014 DARPA established the Biological Technologies Office to oversee basic and applied research in areas including gene editing, neurosciences, and synthetic biology. The office launched the Pandemic Prevention Platform, which helped fund the development of the mRNA technology that is used in the Moderna and Pfizer coronavirus vaccines.
She left the agency in 2017 to move back to California with her family.
“When I left the organization, what was very much on my mind was that the United States has the most powerful innovation engine the world has ever seen,” Prabhakar says. “At the same time, what kept tugging at me was that we have huge aspirations for the future—such as mitigating climate change—that science and technology have to be part of achieving.”
That’s why, in 2019, she helped found Actuate. She served as the nonprofit’s chief executive until 2022, when she took on the role of OSTP director.
Although she didn’t choose her career path because it was her passion, she says, she came to realize that she loves the role that engineering, science, and technology play in the world because of their “power to change how the future unfolds.”
Leading AI regulation worldwide
When Biden asked if Prabhakar would take the OSTP job, she didn’t think twice, she says. “When do you need me to move in?” she says she told him.
“I was so excited to work for the president because he sees science and technology as a necessary part of creating a bright future for the country,” Prabhakar says.
A month after she took office, the generative AI program ChatGPT launched and became a hot topic.
“AI was already being used in different areas, but all of a sudden it became visible to everyone in a way that it really hadn’t been before,” she says.
Regulating AI became a priority for the Biden administration because of the technology’s breadth and power, she says, as well as the rapid pace at which it’s being developed.
Prabhakar led the creation of Biden’s Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence. Signed on 30 October 2022, the order outlines goals such as protecting consumers and their privacy from AI systems, developing watermarking systems for AI-generated content, and warding off intellectual property theft stemming from the use of generative models.
“The executive order is possibly the most important accomplishment in relation to AI,” Prabhakar says. “It’s a tool that mobilizes the [U.S. government’s] executive branch and recognizes that such systems have safety and security risks, but [it] also enables immense opportunity. The order has put the branches of government on a very constructive path toward regulation.”
Meanwhile, the United States spearheaded a U.N. resolution to make regulating AI an international priority. The United Nations adopted the measure this past March. In addition to defining regulations, it seeks to use AI to advance progress on the U.N.’s sustainable development goals.
“There’s much more to be done,” Prabhakar says, “but I’m really happy to see what the president has been able to accomplish, and really proud that I got to help with that.”
-
Video Friday: The Secrets of Shadow Robot’s New Hand
by Evan Ackerman on 09. August 2024. at 15:30
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.
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!
At ICRA 2024, in Tokyo last May, we sat down with the director of Shadow Robot, Rich Walker, to talk about the journey toward developing its newest model. Designed for reinforcement learning, the hand is extremely rugged, has three fingers that act like thumbs, and has fingertips that are highly sensitive to touch.
[ IEEE Spectrum ]
Food Angel is a food delivery robot to help with the problems of food insecurity and homelessness. Utilizing autonomous wheeled robots for this application may seem to be a good approach, especially with a number of successful commercial robotic delivery services. However, besides technical considerations such as range, payload, operation time, autonomy, etc., there are a number of important aspects that still need to be investigated, such as how the general public and the receiving end may feel about using robots for such applications, or human-robot interaction issues such as how to communicate the intent of the robot to the homeless.
[ RoMeLa ]
The UKRI FLF team RoboHike of UCL Computer Science of the Robot Perception and Learning lab with Forestry England demonstrate the ANYmal robot to help preserve the cultural heritage of an historic mine in the Forest of Dean, Gloucestershire, UK.
This clip is from a reboot of the British TV show “Time Team.” If you’re not already a fan of “Time Team,” let me just say that it is one of the greatest retro reality TV shows ever made, where actual archaeologists wander around the United Kingdom and dig stuff up. If they can find anything. Which they often can’t. And also it has Tony Robinson (from “Blackadder”), who runs everywhere for some reason. Go to Time Team Classics on YouTube for 70+ archived episodes.
[ UCL RPL ]
UBTECH humanoid robot Walker S Lite is working in Zeekr’s intelligent factory to complete handling tasks at the loading workstation for 21 consecutive days, and assist its employees with logistics work.
[ UBTECH ]
Current visual navigation systems often treat the environment as static, lacking the ability to adaptively interact with obstacles. This limitation leads to navigation failure when encountering unavoidable obstructions. In response, we introduce IN-Sight, a novel approach to self-supervised path planning, enabling more effective navigation strategies through interaction with obstacles.
[ ETH Zurich paper / IROS 2024 ]
When working on autonomous cars, sometimes it’s best to start small.
[ University of Pennsylvania ]
MIT MechE researchers introduce an approach called SimPLE (Simulation to Pick Localize and placE), a method of precise kitting, or pick and place, in which a robot learns to pick, regrasp, and place objects using the object’s computer-aided design (CAD) model, and all without any prior experience or encounters with the specific objects.
[ MIT ]
Staff, students (and quadruped robots!) from UCL Computer Science wish the Great Britain athletes the best of luck this summer in the Olympic Games & Paralympics.
Walking in tall grass can be hard for robots, because they can’t see the ground that they’re actually stepping on. Here’s a technique to solve that, published in Robotics and Automation Letters last year.
[ ETH Zurich Robotic Systems Lab ]
There is no such thing as excess batter on a corn dog, and there is also no such thing as a defective donut. And apparently, making Kool-Aid drink pouches is harder than it looks.
[ Oxipital AI ]
Unitree has open-sourced its software to teleoperate humanoids in VR for training-data collection.
[ Unitree / GitHub ]
Nothing more satisfying than seeing point-cloud segments wiggle themselves into place, and CSIRO’s Wildcat SLAM does this better than anyone.
[ IEEE Transactions on Robotics ]
A lecture by Mentee Robotics CEO Lior Wolf, on Mentee’s AI approach.
[ Mentee Robotics ]
-
Quantum Cryptography Has Everyone Scrambling
by Margo Anderson on 08. August 2024. at 14:00
While the technology world awaits NIST’s latest “post-quantum” cryptography standards this summer, a parallel effort is underway to also develop cryptosystems that are grounded in quantum technology—what are called quantum-key distribution or QKD systems.
As a result, India, China, and a range of technology organizations in the European Union and United States are researching and developing QKD and weighing standards for the nascent cryptography alternative. And the biggest question of all is how or if QKD fits into a robust, reliable, and fully future-proof cryptography system that will ultimately become the global standard for secure digital communications into the 2030s. As in any emerging technology standard, different players are staking claims on different technologies and implementations of those technologies. And many of the big players are pursuing such divergent options because no technology is a clear winner at the moment.
According to Ciel Qi, a research analyst at the New York-based Rhodium Group, there’s one clear leader in QKD research and development—at least for now. “While China likely holds an advantage in QKD-based cryptography due to its early investment and development, others are catching up,” says Qi.
Two different kinds of “quantum secure” tech
At the center of these varied cryptography efforts is the distinction between QKD and post-quantum cryptography (PQC) systems. QKD is based on quantum physics, which holds that entangled qubits can store their shared information so securely that any effort to uncover it is unavoidably detectable. Sending pairs of entangled-photon qubits to both ends of a network provides the basis for physically secure cryptographic keys that can lock down data packets sent across that network.
Typically, quantum cryptography systems are built around photon sources that chirp out entangled photon pairs—where photon A heading down one length of fiber has a polarization that’s perpendicular to the polarization of photon B heading in the other direction. The recipients of these two photons perform separate measurements that enable both recipients to know that they and only they have the shared information transmitted by these photon pairs. (Otherwise, if a third party had intervened and measured one or both photons first, the delicate photon states would have been irreparably altered before reaching the recipients.)
“People can’t predict theoretically that these PQC algorithms won’t be broken one day.” —Doug Finke, Global Quantum Intelligence
This shared bit the two people on opposite ends of the line have in common then becomes a 0 or 1 in a budding secret key that the two recipients build up by sharing more and more entangled photons. Build up enough shared secret 0s and 1s between sender and receiver, and that secret key can be used for a type of strong cryptography, called a one-time pad, that guarantees a message’s safe transmission and faithful receipt by only the intended recipient.
By contrast, post-quantum cryptography (PQC) is based not around quantum physics but pure math, in which next-generation cryptographic algorithms are designed to run on conventional computers. And it’s the algorithms’ vast complexity that makes PQC security systems practically uncrackable, even by a quantum computer. So NIST—the U.S. National Institute of Standards and Technology—is developing gold-standard PQC systems that will undergird tomorrow’s post-quantum networks and communications.
The big problem with the latter approach, says Doug Finke, chief content officer of the New York-based Global Quantum Intelligence, is PQC is only believed (on very, very good but not infallible evidence) to be uncrackable by a fully-grown quantum computer. PQC, in other words, cannot necessarily offer the ironclad “quantum security” that’s promised.
“People can’t predict theoretically that these PQC algorithms won’t be broken one day,” Finke says. “On the other hand, QKD—there are theoretical arguments based on quantum physics that you can’t break a QKD network.”
That said, real-world QKD implementations might still be hackable via side-channel, device-based, and other clever attacks. Plus, QKD also requires direct access to a quantum-grade fiber optics network and sensitive quantum communications tech, neither of which is exactly commonplace today. “For day-to-day stuff, for me to send my credit card information to Amazon on my cellphone,” Finke says, “I’m not going to use QKD.”
China’s early QKD lead dwindling
According to Qi, China may have originally picked QKD as a focal point of their quantum technology development in part because the U.S. was not directing its efforts that way. “[The] strategic focus on QKD may be driven by China’s desire to secure a unique technological advantage, particularly as the U.S. leads in PQC efforts globally,” she says.
In particular, she points to ramped up efforts to use satellite uplinks and downlinks as the basis for free-space Chinese QKD systems. Citing as a source China’s “father of quantum,” Pan Jianwei, Qi says, “To achieve global quantum network coverage, China is currently developing a medium-high orbit quantum satellite, which is expected to be launched around 2026.”
That said, the limiting factor in all QKD systems to date is their ultimate reliance on a single photon to represent each qubit. Not even the most exquisitely-refined lasers and fiber optic lines can’t escape the vulnerability of individual photons.
QKD repeaters, which would blindly replicate a single photon’s quantum state but not leak any distinguishing information about the individual photons passing through—meaning the repeater would not be hackable by eavesdroppers—do not exist today. But, Finke says, such tech is achievable, though at least 5 to 10 years away. “It definitely is early days,” he says.
“While China likely holds an advantage in QKD-based cryptography due to its early investment and development, others are catching up.” —Ciel Qi, Rhodium Group
“In China they do have a 2,000-kilometer network,” Finke says. “But it uses this thing called trusted nodes. I think they have over 30 in the Beijing to Shanghai network. So maybe every 100 km, they have this unit which basically measures the signal... and then regenerates it. But the trusted node you have to locate on an army base or someplace like that. If someone breaks in there, they can hack into the communications.”
Meanwhile, India has been playing catch-up, according to Satyam Priyadarshy, a senior advisor to Global Quantum Intelligence. Priyadarshy says India’s National Quantum Mission includes plans for QKD communications research—aiming ultimately for QKD networks connecting cities over 2,000-km distances, as well as across similarly long-ranging satellite communications networks.
Priyadarshy points both to government QKD research efforts—including at the Indian Space Research Organization—and private enterprise-based R&D, including by the Bengaluru-based cybersecurity firm QuNu Labs. Priyadarshy says that QuNu, for example, has been working on a hub-and-spoke framework named ChaQra for QKD. (Spectrum also sent requests for comment to officials at India’s Department of Telecommunications, which were unanswered as of press time.)
“A hybrid of QKD and PQC is the most likely solution for a quantum safe network.” —Satyam Priyadarshy, Global Quantum Intelligence
In the U.S. and European Union, similar early-stage efforts are also afoot. Contacted by IEEE Spectrum, officials from the European Telecommunications Standards Institute (ETSI); the International Standards Organization (ISO); the International Electrotechnical Commission (IEC); and the IEEE Communications Society confirmed initiatives and working groups that are now working to both promote QKD technologies and emergent standards now taking shape.
“While ETSI is fortunate to have experts in a broad range of relevant topics, there is a lot to do,” says Martin Ward, senior research scientist based at Toshiba’s Cambridge Research Laboratory in England, and chair of a QKD industry standards group at ETSI.
Multiple sources contacted for this article envisioned a probable future in which PQC will likely be the default standard for most secure communications in a world of pervasive quantum computing. Yet, PQC also cannot avoid its potential Achilles’ heel against increasingly powerful quantum algorithms and machines either. This is where, the sources suggest, QKD could offer the prospect of hybrid secure communications that PQC alone could never provide.
“QKD provides [theoretical] information security, while PQC enables scalab[ility],” Priyadarshy says. “A hybrid of QKD and PQC is the most likely solution for a quantum safe network.” But he added that efforts at investigating hybrid QKD-PQC technologies and standards today are “very limited.”
Then, says Finke, QKD could still have the final say, even in a world where PQC remains preeminent. Developing QKD technology just happens, he points out, to also provide the basis for a future quantum Internet.
“It’s very important to understand that QKD is actually just one use case for a full quantum network,” Finke says.
“There’s a lot of applications, like distributed quantum computing and quantum data centers and quantum sensor networks,” Finke adds. “So even the research that people are doing now in QKD is still very, very helpful because a lot of that same technology can be leveraged for some of these other use cases.”
-
A Non-Engineer’s Journey to IEEE Leadership
by Kathy Pretz on 07. August 2024. at 18:00
Sharlene Brown often accompanied her husband, IEEE Senior Member Damith Wickramanayake, to organization meetings. He has held leadership positions in the IEEE Jamaica Section, in IEEE Region 3, and on the IEEE Member and Geographic Activities board. Both are from Jamaica.
She either waited outside the conference room or helped with tasks such as serving refreshments. Even though her husband encouraged her to sit in on the meetings, she says, she felt uncomfortable doing so because she wasn’t an engineer. Brown is an accountant and human resources professional. Her husband is a computer science professor at the University of Technology, Jamaica, in Kingston. He is currently Region 3’s education activities coordinator and a member of the section’s education and outreach committee for the IEEE Educational Activities Board.
Sharlene Brown
Employer
Maritime Authority of Jamaica, in Kingston
Title
Assistant accountant
Member grade
Senior member
Alma mater
University of Technology, Jamaica, in Kingston; Tsinghua University, in Beijing
After earning her master’s degree in public administration in 2017, Brown says, she felt she finally was qualified to join IEEE, so she applied. Membership is open to individuals who, by education or experience, are competent in different fields including management. She was approved the same year.
“When I joined IEEE, I would spend long hours at night reading various operations manuals and policies because I wanted to know what I was getting into,” she says. “I was always learning. That’s how I got to know a lot of things about the organization.”
Brown is now a senior member and an active IEEE volunteer. She founded the Jamaica Section’s Women in Engineering group; established a student branch; sits on several high-level IEEE boards; and ran several successful recruitment campaigns to increase the number of senior members in Jamaica and throughout Region 3.
Brown was also a member of the subcommittee of the global Women in Engineering committee; she served as membership coordinator and ran several successful senior member campaigns, elevating women on the committee and across IEEE.
Brown also was integral in the promotion and follow-up activities for the One IEEE event held in January at the University of Technology, Jamaica. The first-of-its-kind workshop connected more than 200 participants to each other and to the organization by showcasing Jamaica’s active engineering community. The Jamaica Section has 135 IEEE members.
From factory worker to accountant
Brown grew up in Bog Walk, a rural town in the parish of St. Catherine. Because she had low grades in high school, the only job she was able to get after graduating was as a temporary factory worker at the nearby Nestlé plant. She worked as many shifts as she could to help support her family.
“I didn’t mind working,” she says, “because I was making my mark. Anything I do, I am going to be excellent at, whether it’s cleaning the floor or doing office work.” But she had bigger plans than being a factory worker, she says.
A friend told her about a temporary job overseeing exams at the Jamaican Institute of Management, now part of the University of Technology. Brown worked both jobs for a time until the school hired her full time to do administrative work in its accounting department.
One of the perks of working there was free tuition for employees, and Brown took full advantage. She studied information management and computer applications, Jamaican securities, fraud detection, forensic auditing, and supervisory management, earning an associate degree in business administration in 2007. The school hired her in 2002 as an accountant, and she worked there for five years.
In 2007 she joined the Office of the Prime Minister, in Kingston, initially as an officer handling payments to suppliers. Her hard work and positive attitude got her noticed by other managers, she says. After a month she was tapped by the budget department to become a commitment control officer, responsible for allocating and overseeing funding for four of the country’s ministries.
“What I realized through my volunteer work in IEEE is that you’re never alone. There is always somebody to guide you.”
As a young accountant, she didn’t have hands-on experience with budgeting, but she was a quick learner who produced quality work, she says. She learned the budgeting process by helping her colleagues when her work slowed down and during her lunch breaks.
That knowledge gave her the skills she needed to land her current job as an assistant accountant with the budget and management accounts group in the Maritime Authority of Jamaica accounts department, a position she has held since 2013.
While she was working for the Office of the Prime Minister, Brown continued to further her education. She took night courses at the University of Technology and, in 2012, earned a bachelor’s degree in business administration. She majored in accounting and minored in human resources management.
She secured a full scholarship in 2016 from the Chinese government to study public administration in Beijing at Tsinghua University, earning a master’s degree with distinction in 2017.
Brown says she is now ready to shift to a human resources career. Even though she has been supervising people for more than 17 years, though, she is having a hard time finding an HR position, she says.
Still willing to take on challenges, she is increasing her experience by volunteering with an HR consulting firm in Jamaica. To get more formal training, she is currently working on an HR certification from the Society for Human Resource Management.
Sharlene Brown arranged for the purchase of 350 desk shields for Jamaican schools during the COVID-19 pandemic.Sharlene Brown
Building a vibrant community
After graduating from Tsinghua University, Brown began volunteering for the IEEE Jamaica Section and Region 3.
In 2019 she founded the section’s IEEE Women in Engineering affinity group, which she chaired for three years. She advocated for more women in leadership roles and has run successful campaigns to increase the number of female senior members locally, regionally, and globally across IEEE. She herself was elevated to senior member in 2019.
Brown also got the WIE group more involved in helping the community. One project she is particularly proud of is the purchase of 350 desk shields for Jamaican schools so students could more safely attend classes and examination sessions in person during the COVID-19 pandemic.
Brown was inspired to undertake the project when a student explained on a local news program that his family couldn’t afford Internet for their home, so he was unable to attend classes remotely.
“Every time I watched the video clip, I would cry,” she says. “This young man might be the next engineer, the country’s next minister, or the next professional.
“I’m so happy we were able to get funding from Region 3 and a local organization to provide those shields.”
She established an IEEE student branch at the Caribbean Maritime University, in Kingston. The branch had almost 40 students at the time of formation.
Brown is working to form student branches at other Jamaican universities, and she is attempting to establish an IEEE Power & Energy Society chapter in the section.
She is a member of several IEEE committees including the Election Oversight and Tellers. She serves as chair for the region’s Professional Activities Committee.
“What I realized through my volunteer work in IEEE is that you’re never alone,” she says. “There is always somebody to help guide you. If they don’t know something, they will point you to the person who does.
“Also, you’re allowed to make mistakes,” she says. “In some organizations, if you make a mistake, you might lose your job or have to pay for your error. But IEEE is your professional home, where you learn, grow, and make mistakes.”
On some of the IEEE committees where she serves, she is the only woman of color, but she says she has not faced any discrimination—only respect.
“I feel comfortable and appreciated by the people and the communities I work with,” she says. “That motivates me to continue to do well and to touch lives positively. That’s what makes me so active in serving in IEEE: You’re appreciated and rewarded for your hard work.”
-
Engineering the First Fitbit: The Inside Story
by Tekla S. Perry on 07. August 2024. at 13:00
It was December 2006. Twenty-nine-year-old entrepreneur James Park had just purchased a Wii game system. It included the Wii Nunchuk, a US $29 handheld controller with motion sensors that let game players interact by moving their bodies—swinging at a baseball, say, or boxing with a virtual partner.
Park became obsessed with his Wii.
“I was a tech-gadget geek,” he says. “Anyone holding that nunchuk was fascinated by how it worked. It was the first time that I had seen a compelling consumer use for accelerometers.”
After a while, though, Park spotted a flaw in the Wii: It got you moving, sure, but it trapped you in your living room. What if, he thought, you could take what was cool about the Wii and use it in a gadget that got you out of the house?
The first generation of Fitbit trackers shipped in this package in 2009. NewDealDesign
“That,” says Park, “was the aha moment.” His idea became Fitbit, an activity tracker that has racked up sales of more than 136 million units since its first iteration hit the market in late 2009.
But back to that “aha moment.” Park quickly called his friend and colleague Eric Friedman. In 2002, the two, both computer scientists by training, had started a photo-sharing company called HeyPix, which they sold to CNET in 2005. They were still working for CNET in 2006, but it wasn’t a bad time to think about doing something different.
Friedman loved Park’s idea.
“My mother was an active walker,” Friedman says. “She had a walking group and always had a pedometer with her. And my father worked with augmentative engineering [assistive technology] for the elderly and handicapped. We’d played with accelerometer tech before. So it immediately made sense. We just had to refine it.”
The two left CNET, and in April 2007 they incorporated the startup with Park as CEO and Friedman as chief technology officer. Park and Friedman weren’t trying to build the first step counter—mechanical pedometers date back to the 1960s. They weren’t inventing the first smart activity tracker— BodyMedia, a medical device manufacturer, had in 1999 included accelerometers with other sensors in an armband designed to measure calories burned. And Park and Friedman didn’t get a smart consumer tracker to market first. In 2006, Nike had worked with Apple to launch the Nike+ for runners, a motion-tracking system that required a special shoe and a receiver that plugged into an iPod
Fitbit’s founders James Park [left] and Eric Friedman released their first product in 2009, when this photo was taken. Peter DaSilva/The New York Times/Redux
Park wasn’t aware of any of this when he thought about getting fitness out of the living room, but the two quickly did their research and figured out what they did and didn’t want to do.
“We didn’t want to create something expensive, targeted at athletes,” he says. “Or something that was dumb and not connected to software. And we wanted something that could provide social connection, like photo sharing did.”
That something had to be comfortable to wear all day, be easy to use, upload its data seamlessly so the data could be tracked and shared with friends, and rarely need charging. Not an easy combination of requirements.
“It’s one of those things where the simpler you get, the harder it becomes to design something well,” Park says.
The first Fitbit was designed for women
The first design decision was the biggest one. Where on the body did they expect people to put this wearable? They weren’t going to ask people to buy special shoes, like the Nike+, or wear a thick band on their upper arms, like BodyMedia’s tracker.
They hired NewDealDesign to figure out some of these details.
“In our first two weeks, after multiple discussions with Eric and James, we decided that the project was going to be geared to women,” says Gadi Amit, president and principal designer of NewDealDesign. “That decision was the driver of the form factor.”
“We wanted to start with something familiar to people,” Park says, “and people tended to clip pedometers to their belts.” So a clip-on device made sense. But women generally don’t wear belts.
To do what it needed to do, the clip-on gadget would have to contain a roughly 2.5-by-2.5-centimeter (1-by-1-inch) printed circuit board, Amit recalls. The big breakthrough came when the team decided to separate the electronics and the battery, which in most devices are stacked. “By doing that, and elongating it a bit, we found that women could put it anywhere,” Amit says. “Many would put it in their bras, so we targeted the design to fit a bra in the center front, purchasing dozens of bras for testing.”
The decision to design for women also drove the overall look, to “subdue the user interface,” as Amit puts it. They hid a low-resolution monochrome OLED display behind a continuous plastic cover, with the display lighting up only when you asked it to. This choice helped give the device an impressive battery life.
The earliest Fitbit devices used an animated flower as a progress indicator. NewDealDesign
They also came up with the idea of a flower as a progress indicator—inspired, Park says, by the Tamagotchi, one of the biggest toy fads of the late 1990s. “So we had a little animated flower that would shrink or grow based on how active you were,” Park explains.
And after much discussion over controls, the group gave the original Fitbit just one button.
Hiring an EE—from Dad—to design Fitbit’s circuitry
Park and Friedman knew enough about electronics to build a crude prototype, “stuffing electronics into a box made of cut-up balsam wood,” Park says. But they also knew that they needed to bring in a real electrical engineer to develop the hardware.
Fortunately, they knew just whom to call. Friedman’s father, Mark, had for years been working to develop a device for use in nursing homes, to remotely monitor the position of bed-bound patients. Mark’s partner in this effort was Randy Casciola, an electronics engineer and currently president of Morewood Design Labs.
Eric called his dad, told him about the gadget he and Park envisioned, and asked if he and Casciola could build a prototype.
“Mark and I thought we’d build a quick-and-dirty prototype, something they could get sensor data from and use for developing software. And then they’d go off to Asia and get it miniaturized there,” Casciola recalls. “But one revision led to another.” Casciola ended up working on circuit designs for Fitbits virtually full time until the sale of the company to Google, announced in 2019 and completed in early 2021.
“We saw some pretty scary manufacturers. Dirty facilities, flash marks on their injection-molded plastics, very low precision.”
—James Park“We were just two little guys in a little office in Pittsburgh,” Casciola says. “Before Fitbit came along, we had realized that our nursing-home thing wasn’t likely to ever be a product and had started taking on some consulting work. I had no idea Fitbit would become a household name. I just like working on anything, whether I think it’s a good idea or not, or even whether someone is paying me or not.”
The earliest prototypes were pretty large, about 10 by 15 cm, Casciola says. They were big enough to easily hook up to test equipment, yet small enough to strap on to a willing test subject.
After that, Park and Eric Friedman—along with Casciola, two contracted software engineers, and a mechanical design firm—struggled with turning the bulky prototype into a small and sleek device that counted steps, stored data until it could be uploaded and then transmitted it seamlessly, had a simple user interface, and didn’t need daily charging.
“Figuring out the right balance of battery life, size, and capability kept us occupied for about a year,” Park says.
The Fitbit prototype, sitting on its charger, booted up for the first time in December 2008. James Park
After deciding to include a radio transmitter, they made a big move: They turned away from the Bluetooth standard for wireless communications in favor of the ANT protocol, a technology developed by Garmin that used far less power. That meant the Fitbit wouldn’t be able to upload to computers directly. Instead, the team designed their own base station, which could be left plugged into a computer and would grab data anytime the Fitbit wearer passed within range.
Casciola didn’t have expertise in radio-frequency engineering, so he relied on the supplier of the ANT radio chips: Nordic Semiconductor, in Trondheim, Norway.
“They would do a design review of the circuit board layout,” he explains. “Then we would send our hardware to Norway. They would do RF measurements on it and tell me how to tweak the values of the capacitors and conductors in the RF chain, and I would update the schematic. It’s half engineering and half black magic to get this RF stuff working.”
Another standard they didn’t use was the ubiquitous USB charging connection.
“We couldn’t use USB,” Park says. “It just took up too much volume. Somebody actually said to us, ‘Whatever you do, don’t design a custom charging system because it’ll be a pain, it’ll be super expensive.’ But we went ahead and built one. And it was a pain and super expensive, but I think it added a level of magic. You just plopped your device on [the charger]. It looked beautiful, and it worked consistently.”
Most of the electronics they used were off the shelf, including a 16-bit Texas Instruments MSP430 microprocessor, and 92 kilobytes of flash memory and 4 kb of RAM to hold the operating system, the rest of the code, all the graphics, and at least seven days’ worth of collected data.
The Fitbit was designed to resist sweat, and they generally survived showers and quick dips, says Friedman. “But hot tubs were the bane of our existence. People clipped it to their swimsuits and forgot they had it on when they jumped into the hot tub.”
Fitbit’s demo or die moment
Up to this point, the company was surviving on $400,000 invested by Park, Friedman, and a few people who had backed their previous company. But more money would be needed to ramp up manufacturing. And so a critical next step would be a live public demo, which they scheduled for the TechCrunch conference in San Francisco in September 2008.
Live demonstrations of new technologies are always risky, and this one walked right up to the edge of disaster. The plan was to ask an audience member to call out a number, and then Park, wearing the prototype in its balsa-wood box, would walk that number of steps. The count would sync wirelessly to a laptop projecting to a screen on stage. When Friedman hit refresh on the browser, the step count would appear on the screen. What could go wrong?
A lot. Friedman explains: “You think counting steps is easy, but let’s say you do three steps. One, two, three. When you bring your feet together, is that a step or is that the end? It’s much easier to count 1,000 steps than it is to do 10 steps. If I walk 10 steps and am off by one, that’s a glaring error. With 1,000, that variance becomes noise.”
The first semi-assembled Fitbit records its inaugural step count. James Park
After a lot of practice, the two thought they could pull it off. Then came the demo. “While I was walking, the laptop crashed,” Park says. “I wasn’t aware of that. I was just walking happily. Eric had to reboot everything while I was still walking. But the numbers showed up; I don’t think anyone except Eric realized what had happened.”
That day, some 2,000 preorders poured in. And Fitbit closed a $2 million round of venture investment the next month.
Though Park and Friedman had hoped to get Fitbits into users’ hands—or clipped onto their bras—by Christmas of 2008, they missed that deadline by a year.
The algorithms that determine Fitbit’s count
Part of Fitbit’s challenge of getting from prototype to shippable product was software development. They couldn’t expect users to walk as precisely as Park did for the demo. Instead, the device’s algorithms needed to determine what a step was and what was a different kind of motion—say, someone scratching their nose.
“Data collection was difficult,” Park says. “Initially, it was a lot of us wearing prototype devices doing a variety of different activities. Our head of research, Shelten Yuen, would follow, videotaping so we could go back and count the exact number of steps taken. We would wear multiple devices simultaneously, to compare the data collects against each other.”
Friedman remembers one such outing. “James was tethered to the computer, and he was pretending to walk his dog around the Haight [in San Francisco], narrating this little play that he’s putting on: ‘OK, I’m going to stop. The dog is going to pee on this tree. And now he’s going over there.’ The great thing about San Francisco is that nobody looks strangely at two guys tethered together walking around talking to themselves.”
“Older people tend to have an irregular cadence—to the device, older people look a lot like buses going over potholes.” –James Park
“Pushing baby strollers was an issue,” because the wearer’s arms aren’t swinging, Park says. “So one of our guys put an ET doll in a baby stroller and walked all over the city with it.”
Road noise was another big issue. “Yuen, who was working on the algorithms, was based in Cambridge, Mass.,” Park says. “They have more potholes than we do. When he took the bus, the bus would hit the potholes and [the device would] be bouncing along, registering steps.” They couldn’t just fix the issue by looking for a regular cadence to count steps, he adds, because not everyone has a regular cadence. “Older people tend to have an irregular cadence—to the device, older people look a lot like buses going over potholes.”
Fitbit’s founders enter the world of manufacturing
A consumer gadget means mass manufacturing, potentially in huge quantities. They talked to a lot of contract-manufacturing firms, Park recalls. They realized that as a startup with an unclear future market, they wouldn’t be of interest to the top tier of manufacturers. But they couldn’t go with the lowest-budget operations, because they needed a reasonable level of quality.
“We saw some pretty scary manufacturers,” Park said. “Dirty facilities, flash marks on their injection-molded plastics [a sign of a bad seal or other errors], very low precision.” They eventually found a small manufacturer that was “pretty good but still hungry for business.” The manufacturer was headquartered in Singapore, while their surface-mount supplier, which put components directly onto printed circuit boards, was in Batam, Indonesia.
Workers assemble Fitbits by hand in October of 2008. James Park
Working with that manufacturer, Park and Friedman made some tweaks in the design of the circuitry and the shape of the case. They struggled over how to keep water—and sweat—out of the device, settling on ultrasonic welding for the case and adding a spray-on coating for the circuitry after some devices were returned with corrosion on the electronics. That required tweaking the layout to make sure the coating would get between the chips. The coating on each circuit board had to be checked and touched up by hand. When they realized that the coating increased the height of the chips, they had to tweak the layout some more.
In December 2009, just a week before the ship date, Fitbits began rolling off the production line.
“I was in a hotel room in Singapore testing one of the first fully integrated devices,” Park says. “And it wasn’t syncing to my computer. Then I put the device right next to the base station, and it started to sync. Okay, that’s good, but what was the maximum distance it could sync? And that turned out to be literally just a few inches. In every other test we had done, it was fine. It could sync from 15 or 20 feet [5 or 6 meters] away.”
The problem, Park eventually figured out, occurred when the two halves of the Fitbit case were ultrasonically welded together. In previous syncing tests, the cases had been left unsealed. The sealing process pushed the halves closer together, so that the cable for the display touched or nearly touched the antenna printed on the circuit board, which affected the radio signal. Park tried squeezing the halves together on an unsealed unit and reproduced the problem.
Getting the first generation of Fitbits into mass production required some last-minute troubleshooting. Fitbit cofounder James Park [top, standing in center] helps debug a device at the manufacturer shortly before the product’s 2009 launch. Early units from the production line are shown partially assembled [bottom]. James Park
“I thought, if we could just push that cable away from the antenna, we’d be okay,” Park said. “The only thing I could find in my hotel room to do that was toilet paper. So I rolled up some toilet paper really tight and shoved it in between the cable and the antenna. That seemed to work, though I wasn’t really confident.”
Park went to the factory the next day to discuss the problem—and his solution—with the manufacturing team. They refined his fix—replacing the toilet paper with a tiny slice of foam—and that’s how the first generation of Fitbits shipped.
Fitbit’s fast evolution
The company sold about 5,000 of those $99 first-generation units in 2009, and more than 10 times that number in 2010. The rollout wasn’t entirely smooth. Casciola recalls that Fitbit’s logistics center was sending him a surprising number of corroded devices that had been returned by customers. Casciola’s task was to tear them down and diagnose the problem.
“One of the contacts on the device, over time, was growing a green corrosion,” Casciola says. “But the other two contacts were not.” It turned out the problem came from Casciola’s design of the system-reset trigger, which allowed users to reset the device without a reset button or a removable battery. “Inevitably,” Casciola says, “firmware is going to crash. When you can’t take the battery out, you have to have another way of forcing a reset; you don’t want to have someone waiting six days for the battery to run out before restarting.”
The reset that Casciola designed was “a button on the charging station that you could poke with a paper clip. If you did this with the tracker sitting on the charger, it would reset. Of course, we had to have a way for the tracker to see that signal. When I designed the circuit to allow for that, I ended up with a nominal voltage on one pin.” This low voltage was causing the corrosion.
“If you clipped the tracker onto sweaty clothing—remember, sweat has a high salt content—a very tiny current would flow,” says Casciola. “It was just fractions of a microamp, not enough to cause a reset, but enough, over time, to cause greenish corrosion.”
Cofounders Eric Friedman [left] and James Park visit Fitbit’s manufacturer in December of 2008. James Park
On the 2012 generation of the Fitbit, called the Fitbit One, Casciola added a new type of chip, one that hadn’t been available when he was working on the original design. It allowed the single button to trigger a reset when it was held down for some seconds while the device was sitting on the charger. That eliminated the need for the active pin.
The charging interface was the source of another early problem. In the initial design, the trim of the Fitbit’s plastic casing was painted with chrome. “We originally wanted an actual metal trim,” Friedman says, “but that interfered with the radio signal.”
Chrome wasn’t a great choice either. “It caused problems with the charger interface,” Park adds. “We had to do a lot of work to prevent shorting there.”
They dropped the chrome after some tens of thousands of units were shipped—and then got compliments from purchasers about the new, chrome-less look.
Evolution happened quickly, particularly in the way the device transmitted data. In 2012, when Bluetooth LE became widely available as a new low-power communications standard, the base station was replaced by a small Bluetooth communications dongle. And eventually the dongles disappeared altogether.
“We had a huge debate about whether or not to keep shipping that dongle,” Park says. “Its cost was significant, and if you had a recent iPhone, you didn’t need it. But we didn’t want someone buying the device and then returning it because their cellphone couldn’t connect.” The team closely tracked the penetration rate of Bluetooth LE in cellphones; when they felt that number was high enough, they killed off the dongle.
Fitbit’s wrist-ward migration
After several iterations of the original Fitbit design, sometimes called the “clip” for its shape, the fitness tracker moved to the wrist. This wasn’t a matter of simply redesigning the way the device attached to the body but a rethinking of algorithms.
The impetus came from some users’ desire to better track their sleep. The Fitbit’s algorithms allowed it to identify sleep patterns, a design choice that, Park says, “was pivotal, because it changed the device from being just an activity tracker to an all-day wellness tracker.” But nightclothes didn’t offer obvious spots for attachment. So the Fitbit shipped with a thin fabric wristband intended for use just at night. Users began asking customer support if they could keep the wristband on around the clock. The answer was no; Fitbit’s step-counting algorithms at the time didn’t support that.
“My father, who turned 80 on July 5, is fixated on his step count. From 11 at night until midnight, he’s in the parking garage, going up flights of stairs. And he is in better shape than I ever remember him.” —Eric Friedman
Meanwhile, a cultural phenomenon was underway. In the mid-2000s, yellow Livestrong bracelets, made out of silicone and sold to support cancer research, were suddenly everywhere. Other causes and movements jumped on the trend with their own brightly colored wristbands. By early 2013, Fitbit and its competitors Nike and Jawbone had launched wrist-worn fitness trackers in roughly the same style as those trendy bracelets. Fitbit’s version was called the Flex, once again designed by NewDealDesign.
A no-button user interface for the Fitbit Flex
The Flex’s interface was even simpler than the original Fitbit’s one button and OLED screen: It had no buttons and no screen, just five LEDs arranged in a row and a vibrating motor. To change modes, you tapped on the surface.
“We didn’t want to replace people’s watches,” Park says. The technology wasn’t yet ready to “build a compelling device—one that had a big screen and the compute power to drive really amazing interactions on the wrist that would be worthy of that screen. The technology trends didn’t converge to make that possible until 2014 or 2015.”
The Fitbit Flex [right], the first Fitbit designed to be worn on the wrist, was released in 2013. It had no buttons and no screen. Users controlled it by tapping; five LEDs indicated progress toward a step count selected via an app [left]. iStock
“The amount of stuff the team was able to convey with just the LEDs was amazing,” Friedman recalls. “The status of where you are towards reaching your [step] goal, that’s obvious. But [also] the lights cycling to show that it’s searching for something, the vibrating when you hit your step goal, things like that.”
The tap part of the interface, though, was “possibly something we didn’t get entirely right,” Park concedes. It took much fine-tuning of algorithms after the launch to better sort out what was not tapping—like applauding. Even more important, some users couldn’t quite intuit the right way to tap.
“If it works for 98 percent of your users, but you’re growing to millions of users, 2 percent really starts adding up,” Park says. They brought the button back for the next generation of Fitbit devices.
And the rest is history
In 2010, its first full year on the market, the Fitbit sold some 50,000 units. Fitbit sales peaked in 2015, with almost 23 million devices sold that year, according to Statista. Since then, there’s been a bit of a drop-off, as multifunctional smart watches have come down in price and grown in popularity and Fitbit knockoffs entered the market. In 2021, Fitbit still boasted more than 31 million active users, according to Market.us.Media. And Fitbit may now be riding the trend back to simplicity, as people find themselves wanting to get rid of distractions and move back to simpler devices. I see this happening in my own family: My smartwatch-wearing daughter traded in that wearable for a Fitbit Charge 6 earlier this year.
Related Articles
The Consumer Electronics Hall of Fame: Fitbit
Fitbit went public in 2015 at a valuation of $4.1 billion. In 2021 Google completed its $2.1 billion purchase of the company and absorbed it into its hardware division. In April of this year, Park and Friedman left Google. Early retirement? Hardly. The two, now age 47, have started a new company that’s currently in stealth mode.
The idea of encouraging people to be active by electronically tracking steps has had staying power.
“My father, who turned 80 on July 5, is fixated on his step count,” Friedman says. “From 11 at night until midnight, he’s in the parking garage, going up flights of stairs. And he is in better shape than I ever remember him.”
What could be a better reward than that?
This article appears in the September 2024 print issue.
-
Fitting It All In: Keys to Mastering Work-Life Balance
by Mark Wehde on 06. August 2024. at 18:00
This article is part of our exclusive career advice series in partnership with the IEEE Technology and Engineering Management Society.
With technological advancement and changing societal expectations, the concept of work-life balance has become an elusive goal for many, particularly within the engineering community. The drive to remain continuously engaged with work, the pressure to achieve perfection, and the challenge of juggling work and personal responsibilities have created a landscape where professional and personal spheres are in constant negotiation.
This article covers several factors that can disrupt work-life balance, with recommendations on how to address them.
The myth of urgency
In an era dominated by instant communication via email and text messages, the expectation to respond quickly has led to an illusion of urgency. The perpetual state of constant alertness blurs the distinction between what’s urgent and what isn’t.
Recognizing that not every email message warrants an immediate response is the first step in deciding what’s important. By prioritizing responses based on actual importance, individuals can reclaim control over their time, reduce stress, and foster a more manageable workload.
Throughout my career, I have found that setting specific times to check and respond to email helps avoid distractions throughout the day. There are programs that prioritize email and classify tasks based on its urgency and importance.
Another suggestion is to unsubscribe from unnecessary newsletters and set up filters that move unwanted email to a specific folder or the trash before it reaches your inbox.
Cutting back the endless workday
Today’s work environment, characterized by remote access and flexible hours, has extended the workday beyond a set schedule and has encroached on personal time. The situation is particularly prevalent among engineers committed to solving complex problems, leading to a scenario where work is a constant companion—which leaves little room for personal pursuits or time with family.
A balanced life is healthier and more sustainable, and it enriches the quality of our work and our relationships with those we love.
Establishing clear boundaries between work and personal time is essential. One way to do so is to communicate clear working hours to your manager, coworkers, and clients. You can use tools such as email autoresponders and do-not-disturb modes to reinforce your boundaries.
It’s important to recognize that work, while integral, is only one aspect of life.
The quest for perfectionism
The pursuit of perfection is a common trap for many professionals, leading to endless revisions and dissatisfaction with one’s work. The quest not only wastes an inordinate amount of time. It also detracts from the quality of life.
Embracing the philosophy that “it doesn’t have to be perfect” can liberate individuals from the trap. By aiming for excellence rather than perfection, one can achieve high standards of work while also making time for personal growth and happiness.
To help adopt such a mindset, practice setting realistic standards for different tasks by asking yourself what level of quality is truly necessary for each. Allocating a fixed amount of time to specific tasks can help prevent endless tweaking.
The necessity of exercise
Physical activity often takes a back seat to busy schedules and is often viewed as negotiable or secondary to work and family responsibilities. Exercise, however, is a critical component for maintaining mental and physical health. Integrating regular physical activity into one’s routine is not just beneficial; it’s essential for maintaining balance and enhancing your quality of life.
One way to ensure you are taking care of your health is to schedule exercise as a nonnegotiable activity in your calendar, similar to important meetings or activities. Also consider integrating physical activity into your daily routine, such as riding a bicycle to work, walking to meetings, and taking short strolls around your office building. If you work from home, take a walk around your neighborhood.
Sleep boosts productivity
Contrary to the glorification of overwork and sleep deprivation in some professional circles, sleep is a paramount factor in maintaining high levels of productivity and creativity. Numerous studies have shown that adequate sleep—seven to nine hours for most adults—enhances cognitive functions, problem-solving skills, and memory retention.
For engineers and others in professions where innovation and precision are paramount, neglecting sleep can diminish the quality of work and the capacity for critical thinking.
Sleep deprivation has been linked to a variety of health issues including increased risk of cardiovascular disease, diabetes, and stress-related conditions.
Prioritizing sleep is not a luxury but a necessity for those aiming to excel in their career while also enjoying a fulfilling personal life.
Begin your bedtime routine at the same time each night to cue your body that it’s time to wind down. For a smooth transition to sleep, try adjusting lighting, reducing noise, and engaging in relaxing activities such as reading or listening to calm music.
Relaxation is the counterbalance to stress
Relaxation is crucial for counteracting the effects of stress and preventing burnout. Techniques such as meditation, deep-breathing exercises, yoga, and engaging in leisure activities that bring joy can significantly reduce stress levels, thereby enhancing emotional equilibrium and resilience.
Spending time with friends and family is another effective relaxation strategy. Social interactions with loved ones can provide emotional support, happiness, and a sense of belonging, all of which are essential for limiting stress and promoting mental health. The social connections help build a support network that can serve as a buffer against life’s challenges, providing a sense of stability and comfort.
Allow yourself to recharge and foster a sense of fulfillment by allocating time each week to pursue interests that enrich your life. Also consider incorporating relaxation techniques in your daily routine, such as mindfulness meditation or short walks outdoors.
Guarding time and energy
In the quest for balance, learning to say no and ruthlessly eliminating activities that do not add value are invaluable skills. Make conscious choices about how to spend your time and energy, focusing on activities that align with personal and professional priorities. By doing so, individuals can protect their time, reduce stress, and dedicate themselves more fully to meaningful pursuits.
Practice assertiveness in communicating your capacity and boundaries to others. When asked to take on an additional task, it’s important to consider the impact on your current priorities. Don’t hesitate to decline politely if the new task doesn’t align.
Challenges for women
When discussing work-life balance, it’s essential to acknowledge the specific challenges faced by women, particularly in engineering. They are often expected to manage household duties, childcare, and their professional responsibilities while also supporting their partner’s career goals.
It can be especially challenging for women who strive to meet high standards at work and home. Recognizing and addressing their challenges is crucial in fostering an environment that supports balance for everyone.
One way to do that is to have open discussions with employers about the challenges and the support needed in the workplace and at home. Advocating for company policies that support work-life balance, such as a flexible work schedule and parental leave, is important.
Achieving a healthy work-life balance in the engineering profession—and indeed in any high-pressure field—is an ongoing process that requires self-awareness, clear priorities, and the courage to set boundaries.
It involves a collective effort by employers and workers to recognize the value of balance and to create a culture that supports it.
By acknowledging the illusion of constant urgency, understanding our limitations, and addressing the particular challenges faced by women, we can move toward a future where professional success and personal fulfillment are mutually reinforcing.
A balanced life is healthier and more sustainable, and it enriches the quality of our work and our relationships with those we love.
-
Figure 02 Robot Is a Sleeker, Smarter Humanoid
by Evan Ackerman on 06. August 2024. at 13:06
Today, Figure is introducing the newest, slimmest, shiniest, and least creatively named next generation of its humanoid robot: Figure 02. According to the press release, Figure 02 is the result of “a ground-up hardware and software redesign” and is “the highest performing humanoid robot,” which may even be true for some arbitrary value of “performing.” Also notable is that Figure has been actively testing robots with BMW at a manufacturing plant in Spartanburg, S.C., where the new humanoid has been performing “data collection and use case training.”
The rest of the press release is pretty much, “Hey, check out our new robot!” And you’ll get all of the content in the release by watching the videos. What you won’t get from the videos is any additional info about the robot. But we sent along some questions to Figure about these videos, and have a few answers from Michael Rose, director of controls, and Vadim Chernyak, director of hardware.
First, the trailer:
How many parts does Figure 02 have, and is this all of them?
Figure: A couple hundred unique parts and a couple thousand parts total. No, this is not all of them.
Does Figure 02 make little Figure logos with every step?
Figure: If the surface is soft enough, yes.
Swappable legs! Was that hard to do, or easier to do because you only have to make one leg? Figure: We chose to make swappable legs to help with manufacturing.
Is the battery pack swappable too?
Figure: Our battery is swappable, but it is not a quick swap procedure.
What’s that squishy-looking stuff on the back of Figure 02’s knees and in its elbow joints?
Figure: These are soft stops which limit the range of motion in a controlled way and prevent robot pinch points
Where’d you hide that thumb motor?
Figure: The thumb is now fully contained in the hand.
Tell me about the “skin” on the neck!
Figure: The skin is a soft fabric which is able to keep a clean seamless look even as the robot moves its head.
And here’s the reveal video:
When Figure 02’s head turns, its body turns too, and its arms move. Is that necessary, or aesthetic?
Figure: Aesthetic.
The upper torso and shoulders seem very narrow compared to other humanoids. Why is that?
Figure: We find it essential to package the robot to be of similar proportions to a human. This allows us to complete our target use cases and fit into our environment more easily.
What can you tell me about Figure 02’s walking gait?
Figure: The robot is using a model predictive controller to determine footstep locations and forces required to maintain balance and follow the desired robot trajectory.
How much runtime do you get from 2.25 kilowatt-hours doing the kinds of tasks that we see in the video?
Figure: We are targeting a 5-hour run time for our product.
Slick, but also a little sinister?Figure
This thing looks slick. I’d say that it’s maybe a little too far on the sinister side for a robot intended to work around humans, but the industrial design is badass and the packaging is excellent, with the vast majority of the wiring now integrated within the robot’s skins and flexible materials covering joints that are typically left bare. Figure, if you remember, raised a US $675 million Series B that valued the company at $2.6 billion, and somehow the look of this robot seems appropriate to that.
I do still have some questions about Figure 02, such as where the interesting foot design came from and whether a 16-degree-of-freedom hand is really worth it in the near term. It’s also worth mentioning that Figure seems to have a fair number of Figure 02 robots running around—at least five units at its California headquarters, plus potentially a couple of more at the BMW Spartanburg manufacturing facility.
I also want to highlight this boilerplate at the end of the release: “our humanoid is designed to perform human-like tasks within the workforce and in the home.” We are very, very far away from a humanoid robot in the home, but I appreciate that it’s still an explicit goal that Figure is trying to achieve. Because I want one.
-
Build a Radar Cat Detector
by Stephen Cass on 29. July 2024. at 14:00
You have a closed box. There may be a live cat inside, but you won’t know until you open the box. For most people, this situation is a theoretical conundrum that probes the foundations of quantum mechanics. For me, however, it’s a pressing practical problem, not least because physics completely skates over the vital issue of how annoyed the cat will be when the box is opened. But fortunately, engineering comes to the rescue, in the form of a new US $50 maker-friendly pulsed coherent radar sensor from SparkFun.
Perhaps I should back up a little bit. Working from home during the pandemic, my wife and I discovered a colony of feral cats living in the backyards of our block in New York City. We reversed the colony’s growth by doing trap-neuter-return (TNR) on as many of its members as we could, and we purchased three Feralvilla outdoor shelters to see our furry neighbors through the harsh New York winters. These roughly cube-shaped insulated shelters allow the cats to enter via an opening in a raised floor. A removable lid on top allows us to replace straw bedding every few months. It’s impossible to see inside the shelter without removing the lid, meaning you run the risk of surprising a clawed predator that, just moments before, had been enjoying a quiet snooze.
The enclosure for the radar [left column] is made of basswood (adding cat ears on top is optional). A microcontroller [top row, middle column] processes the results from the radar module [top row, right column] and illuminates the LEDs [right column, second from top] accordingly. A battery and on/off switch [bottom row, left to right] make up the power supply.James Provost
Feral cats respond to humans differently than socialized pet cats do. They see us as threats rather than bumbling servants. Even after years of daily feeding, most of the cats in our block’s colony will not let us approach closer than a meter or two, let alone suffer being touched. They have claws that have never seen a clipper. And they don’t like being surprised or feeling hemmed in. So I wanted a way to find out if a shelter was occupied before I popped open its lid for maintenance. And that’s where radar comes in.
SparkFun’s pulsed coherent radar module is based on Acconeer’s low-cost A121 sensor. Smaller than a fingernail, the sensor operates at 60 gigahertz, which means its signal can penetrate many common materials. As the signal passes through a material, some of it is reflected back to the sensor, allowing you to determine distances to multiple surfaces with millimeter-level precision. The radar can be put into a “presence detector” mode—intended to flag whether or not a human is present—in which it looks for changes in the distance of reflections to identify motion.
As soon as I saw the announcement for SparkFun’s module, the wheels began turning. If the radar could detect a human, why not a feline? Sure, I could have solved my is-there-a-cat-in-the-box problem with less sophisticated technology, by, say, putting a pressure sensor inside the shelter. But that would have required a permanent setup complete with weatherproofing, power, and some way of getting data out. Plus I’d have to perform three installations, one for each shelter. For information I needed only once every few months, that seemed a bit much. So I ordered the radar module, along with a $30 IoT RedBoard microcontroller. The RedBoard operates at the same 3.3 volts as the radar and can configure the module and parse its output.
If the radar could detect a human, why not a feline?
Connecting the radar to the RedBoard was a breeze, as they both have Qwiic 4-wire interfaces, which provides power along with an I2C serial connection to peripherals. SparkFun’s Arduino libraries and example code let me quickly test the idea’s feasibility by connecting the microcontroller to a host computer via USB, and I could view the results from the radar via a serial monitor. Experiments with our indoor cats (two defections from the colony) showed that the motion of their breathing was enough to trigger the presence detector, even when they were sound asleep. Further testing showed the radar could penetrate the wooden walls of the shelters and the insulated lining.
The next step was to make the thing portable. I added a small $11 lithium battery and spliced an on/off switch into its power lead. I hooked up two gumdrop LEDs to the RedBoard’s input/output pins and modified SparkFun’s sample scripts to illuminate the LEDs based on the output of the presence detector: a green LED for “no cat” and red for “cat.” I built an enclosure out of basswood, mounted the circuit boards and battery, and cut a hole in the back as a window for the radar module. (Side note: Along with tending feral cats, another thing I tried during the pandemic was 3D-printing plastic enclosures for projects. But I discovered that cutting, drilling, and gluing wood was faster, sturdier, and much more forgiving when making one-offs or prototypes.)
The radar sensor sends out 60-gigahertz pulses through the walls and lining of the shelter. As the radar penetrates the layers, some radiation is reflected back to the sensor, which it detects to determine distances. Some materials will reflect the pulse more strongly than others, depending on their electrical permittivity. James Provost
I also modified the scripts to adjust the range over which the presence detector scans. When I hold the detector against the wall of a shelter, it looks only at reflections coming from the space inside that wall and the opposite side, a distance of about 50 centimeters. As all the cats in the colony are adults, they take up enough of a shelter’s volume to intersect any such radar beam, as long as I don’t place the detector near a corner.
I performed in-shelter tests of the portable detector with one of our indoor cats, bribed with treats to sit in the open box for several seconds at a time. The detector did successfully spot him whenever he was inside, although it is prone to false positives. I will be trying to reduce these errors by adjusting the plethora of available configuration settings for the radar. But in the meantime, false positives are much more desirable than false negatives: A “no cat” light means it’s definitely safe to open the shelter lid, and my nerves (and the cats’) are the better for it.
-
Andrew Ng: Unbiggen AI
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.
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 AII 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.”
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 NgFor 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.
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 NgSynthetic 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.
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.
This article appears in the April 2022 print issue as “Andrew Ng, AI Minimalist.”
-
How AI Will Change Chip Design
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.
-
Atomically Thin Materials Significantly Shrink Qubits
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.