+ What is an RSS feed?
Current feed - IEEE Spectrum
For anyone working in today’s rapidly evolving science, technology, engineering, and mathematics fields, visibility, authenticity, and connection are no longer optional; they are essential. But there is a lack of resources for STEM professionals, especially women, looking to express themselves fully, build meaningful networks, and lead with confidence.
To help, IEEE Women in Engineering (WIE) recently launched a podcast series in which experts from around the world inspire and inform to ignite change.
The series aims to amplify the diverse experiences of women from STEM fields. Through candid conversations and expert insights, the podcast goes beyond technical talks to explore the human side of innovation, navigating burnout, balancing career ambition with well-being, and building successful, sustainable careers.
The series is a volunteer and staff-run initiative.
“In the early days of planning, our vision was just a spark shared among passionate volunteers eager to shape each episode and guest experience,” says Geetika Tandon, cochair of the IEEE WIE podcast subcommittee. “Seeing our podcast grow from those first conversations into a vibrant reality has been truly rewarding. We can’t wait for it to expand further.”
“I’m excited that we’ve brought the drawings on our whiteboard and day planners to life,” says Kelly Onu, who is also cochair.
New episodes are released on the third Wednesday of each month.
The podcast’s premier episode, “Moms Who Innovate,” which debuted in May, features candid conversations with two executive coaches, authors, and TEDx speakers. Adaeze Iloeje-Udeogalanya, is the founder of African Women in STEM, which provides education, mentoring, and networking opportunities. Cassie Leonard is a seasoned aerospace professional who founded ELMM Coaching. Leonard offers one-on-one advice for professionals looking to grow their career and achieve a better work-life balance. She authored STEM Moms: Design, Build, and Test to Create the Work-Life of Your Dreams, a book that guides women by drawing from her experiences as a working mother.
Onu, who moderated the episode, spoke with Iloeje-Udeogalanya and Leonard about the ebb and flow of being a mother while building a career. Both guests described how their background as engineers shaped the way they approach motherhood and community. They emphasized the importance of creating a support system that makes the busier times of life more manageable.
Leonard said she “engineered her neighborhood” and shares the responsibilities of dropping off children at school, babysitting after school, and other day-to-day tasks.
“As the podcast series grows, our mission is to shine a spotlight on the real-life adventures (and occasional misadventures) of women in STEM. We want to share late-night brainstorms, coffee-fueled breakthroughs, and the moment when someone finally figures out how to unmute themselves on virtual meeting platforms.” —Geetika Tandon
Innovation for moms isn’t only about professional success, the duo said, but also about designing the kind of community that helps them thrive.
The June episode, “Global Perspectives on Women in STEM,” led by Tandon, offered practical strategies for navigating work-life-balance challenges. Together with guest Sanyogita Shamsunder, CTO of telecommunications company GeoLinks in San Francisco, Tandon explored different perspectives of women around the world.
Rawan Alghamdi, a wireless communication researcher at the King Abdullah University of Science and Technology, in Saudi Arabia, and an IEEE graduate student member hosted August’s episode, “PIE Framework: Presence, Image, and Exposure for Professionals in STEM.” Alghamdi spoke with Jahnavi Brenner, an executive coach and former engineer, who explained the PIE model, which challenges the long-held belief that technical skills alone are enough to advance one’s career.
Brenner said professionals must strategically build an authentic personal brand to dictate how they are perceived by colleagues and how visible they are within their networks and industry. She said it is especially vital for women and underrepresented groups, who often face systemic barriers to recognition and promotion.
October’s episode, “Balancing Work and Life in STEM Careers,” tackled struggles parents face raising a family while working full time. It was moderated by Abinaya Inbamani, a mentor who has contributed to the successful deployment of IoT systems used for smart health care, renewable energy, and cybersecurity.
She covered the intense logistics and emotional toll of balancing a demanding career with the responsibilities of parenthood.
Listeners also learned time-management strategies and boundary-setting techniques, such as reframing guilt as a reminder of care and responsibility rather than failure; accepting that it’s all right to procrastinate occasionally rather than push through unhealthy stress; and organizing the day with clear boundaries between work and home.
“We don’t have to do it all,” Inbamani said. “Sometimes balance is simply choosing what matters most in that moment.”
Upcoming episodes will focus on being present parents, setting boundaries in high-pressure environments, and redefining success on one’s own terms, Tandon and Onu say.
In the works is an episode spotlighting tech trailblazer Nimisha Morkonda Gnanasekaran, who was recognized by the IEEE Computer Society as one of its Top 30 Early Career Professionals this year. She is the director of data science and advanced analytics at Western Digital, based in San Jose, Calif.
Another episode, Tandon and Onu say, will feature a conversation with Cynthia Kane, author of The Pause Principle: How to Keep Your Cool in Tough Situations, on navigating difficult workplace conversations without shutting down or losing one’s temper. The episode will tackle critical issues and career struggles women face, Tandon and Onu say. A study that found as many as 50 percent of women leave their STEM career within five years.
IEEE WIE is seeing the impact the podcast is having on listeners. Several say they tune in not just for advice but also to connect with others. Others say the podcast makes them feel they are not alone in their challenges or career aspirations.
The majority of listeners are in Canada, India, Japan, Saudi Arabia, Türkiye, and the United States. Onu says she hopes the audience expands to include more countries.
“I hope this podcast hops across continents, sneaks into earbuds everywhere, and becomes a trusty sidekick in women’s STEM journeys—cheering them on as they conquer equations, break barriers, and maybe even invent a robot that makes perfect coffee,” Tandon says. “As the podcast series grows, our mission is to shine a spotlight on the real-life adventures (and occasional misadventures) of women in STEM. We want to share late-night brainstorms, coffee-fueled breakthroughs, and the moment when someone finally figures out how to unmute themselves on virtual meeting platforms.”
Through personal tales, inspiring journeys, and a parade of trailblazing leaders who have tackled obstacles, IEEE WIE is celebrating the grit, wit, and brilliance of women in STEM.
Whether you’re a student just beginning your STEM journey, a mid-career professional seeking clarity, or a leader looking to give back to your profession, the podcast offers a space to learn, reflect, and rise together.
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.
Enjoy today’s videos!
Unlike existing hybrid designs, Duawlfin eliminates the need for additional actuators or propeller-driven ground propulsion by leveraging only its standard quadrotor motors and introducing a differential drivetrain with one-way bearings. The seamless transitions between aerial and ground modes further underscore the practicality and effectiveness of our approach for applications like urban logistics and indoor navigation.
[ HiPeR Lab ]
I appreciate the softness of NEO’s design, but those fingers look awfully fragile.
[ 1X ]
Imagine reaching into your backpack to find your keys. Your eyes guide your hand to the opening, but once inside, you rely almost entirely on touch to distinguish your keys from your wallet, phone, and other items. This seamless transition between sensory modalities (knowing when to rely on vision versus touch) is something humans do effortlessly but robots struggle with. The challenge isn’t just about having multiple sensors. Modern robots are equipped with cameras, tactile sensors, depth sensors, and more. The real problem is **how to integrate these different sensory streams**, especially when some sensors provide sparse but critical information at key moments. Our solution comes from rethinking how we combine modalities. Instead of forcing all sensors through a single network, we train separate expert policies for each modality and learn how to combine their action predictions at the policy level.
Multi-university Collaboration presented via [ GitHub ]
Thanks, Haonan!
Happy (somewhat late) Halloween from Pollen Robotics!
[ Pollen Robotics ]
In collaboration with our colleagues from Iowa State and University of Georgia, we have put our pipe-crawling worm robot to test in the field. See it crawls through corrugated drainage pipes in a stream, and a smooth section of a subsurface drainage system.
[ Paper ] from [ Smart Microsystems Laboratory, Michigan State University ]
Heterogeneous robot teams operating in realistic settings often must accomplish complex missions requiring collaboration and adaptation to information acquired online. Because robot teams frequently operate in unstructured environments — uncertain, open-world settings without prior maps — subtasks must be grounded in robot capabilities and the physical world. We present SPINE-HT, a framework that addresses these limitations by grounding the reasoning abilities of LLMs in the context of a heterogeneous robot team through a three-stage process. In real-world experiments with a Clearpath Jackal, a Clearpath Husky, a Boston Dynamics Spot, and a high-altitude UAV, our method achieves an 87% success rate in missions requiring reasoning about robot capabilities and refining subtasks with online feedback.
[ SPINE-HT ] from [ GRASP Lab, University of Pennsylvania ]
Astribot keeping itself busy at IROS 2025.
[ Astribot ]
In two papers published in Matter and Advanced Science, a team of scientists from the Physical Intelligence Department at the Max Planck Institute for Intelligent Systems in Stuttgart, Germany, developed control strategies for influencing the motion of self-propelling oil droplets. These oil droplets mimic single-celled microorganisms and can autonomously solve a complex maze by following chemical gradients. However, it is very challenging to integrate external perturbation and use these droplets in robotics. To address these challenges, the team developed magnetic droplets that still possess life-like properties and can be controlled by external magnetic fields. In their work, the researchers showed that they are able to guide the droplet’s motion and use them in microrobotic applications such as cargo transportation.
Everyone has fantasized about having an embodied avatar! Full-body teleoperation and full-body data acquisition platform is waiting for you to try it out!
[ Unitree ]
It’s not a humanoid, but it right now safely does useful things and probably doesn’t cost all that much to buy or run.
[ Naver Labs ]
This paper presents a curriculum-based reinforcement learning framework for training precise and high-performance jumping policies for the robot `Olympus’. Separate policies are developed for vertical and horizontal jumps, leveraging a simple yet effective strategy. Experimental validation demonstrates horizontal jumps up to 1.25 m with centimeter accuracy and vertical jumps up to 1.0 m. Additionally, we show that with only minor modifications, the proposed method can be used to learn omnidirectional jumping.
[ Paper ] from [ Autonomous Robots Lab, Norwegian University of Science and Technology ]
Heavy payloads are no problem for it: The new KR TITAN ultra moves payloads of up to 1500 kg, making the heavy lifting extreme in the KUKA portfolio.
[ Kuka ]
Good luck getting all of the sand out of that robot. Perhaps a nice oil bath is in order?
[ DEEP Robotics ]
This CMU RI Seminar is from Yuke Zhu at University of Texas at Austin, on “Toward Generalist Humanoid Robots: Recent Advances, Opportunities, and Challenges.”
In an era of rapid AI progress, leveraging accelerated computing and big data has unlocked new possibilities to develop generalist AI models. As AI systems like ChatGPT showcase remarkable performance in the digital realm, we are compelled to ask: Can we achieve similar breakthroughs in the physical world — to create generalist humanoid robots capable of performing everyday tasks? In this talk, I will outline our data-centric research principles and approaches for building general-purpose robot autonomy in the open world. I will present our recent work leveraging real-world, synthetic, and web data to train foundation models for humanoid robots. Furthermore, I will discuss the opportunities and challenges of building the next generation of intelligent robots.
[ Carnegie Mellon University Robotics Institute ]
A new onboard system allows ocean-going vessels to share real-time sea condition data, giving crews early warnings and helping them navigate more safely. The system will analyze data related to navigation, vessel behavior, and the environment to give ship crews guidance at sea.
While casualties from ship collisions and groundings have declined, the overall number of maritime incidents are on the rise, up 22 percent in recent years, driven by aging vessels and equipment failures.
Orca AI, a London-based autonomous maritime navigation company, has introduced a software feature called Co-Captain, aiming to reduce those incidents. Co-Captain is an addition to the company’s existing SeaPod real-time decision support system, which bridge officers can use while at sea to navigate better.
Co-Captain provides information about severe weather, including recommendations to specific ships based on their size and shape.Orca AI
“Co-Captain is a network of vessels using Orca to capture events worldwide and share insights. Think of it like the navigation app you use in your car: it tells you about traffic or roadblocks in advance so you can adjust your route,” says Yarden Gross, the CEO and co-founder of Orca AI.
Gross says that Co-Captain frequently collects data from sensors on board vessels and sends it to the cloud to improve ship performance and safety for vessels globally.
OrcaAI, founded in 2018 by Gross and Dor Raviv, the CTO, began with SeaPod and Fleet View. While SeaPod collects and analyzes data on individual ships, Fleet View gathers that data in the cloud to give fleet managers on shore better visibility into larger operations.
Co-Captain integrates with the existing system to provide proactive insights to improve fleet performance and safety. Today, ship officers rely on tools like radar, the automatic identification system (AIS), and the Electronic Chart Display and Information System (ECDIS) monitor the positions of other vessels and avoid collisions, but much of the work remains manual.
Co-Captain identifies various navigational hazards to a ship’s crew. The crew can also manually tag obstacles or other concerns.Orca AI
Gross described Co-Captain as the next generation of AIS, the network that transmits basic information like a ship’s position, name, and heading over very high frequency (VHF) signals ranging from 30 to 300 megahertz. Unlike AIS, which tracks only a ship’s position, Co-Captain also monitors onboard conditions. For example, if a ship reports a pitch of 3 degrees and a roll of 5 degrees in rough seas, Co-Captain uses that data to anticipate how current conditions will impact nearby ships, adjusted for their size and design. Co-Captain then sends tailored recommendations to those vessels’ crews.
“Every ship acts as a node in a larger network, and each node—the vessel itself—has an onboard AI platform. This platform collects data from multiple sensors in real time,” Gross says. Using cameras and computer vision, the AI model can detect bad weather, low visibility, tall waves, or strong winds, then the platform analyzes the data to provide tailored guidance.
All data is anonymized. Gross says that a ship’s movements, timing, or route can reveal valuable information. “By anonymizing the data, Co-Captain can share critical safety alerts such as GPS interference, severe weather, or high traffic without ever exposing which vessel reported it or where it came from.”
Gross says that Orca AI is working on integrating Co-Captain with more bridge systems, such as Navigational Telex (NAVTEX) and ECDIS, so that relevant alerts and updates are centralized.
The company’s long-term goal is to provide real-time notifications focused on the most important events along a ship’s route, giving captains information they can act on quickly to support safer and more efficient operations. The platform is already in use on over 1,200 vessels.
In Menifee, Calif., six newly built homes are testing a first for North America: electric vehicles that can power houses through the Combined Charging System (CCS) high-power DC charging standard. Each home uses a host Kia EV9 electric vehicle connected to a Wallbox Quasar 2 bidirectional charger, allowing the car’s 100-kilowatt-hour (kWh) battery to run essential circuits during blackouts or periods when electricity prices are high. The setup is the first residential vehicle-to-home (V2H) system in the United States that uses the Combined Charging System (CCS) standard. The CCS is the charging system commonly used in European and North American residential and public charging facilities.
Since July, the homes’ smart electrical panels have automatically managed two-way power flow—charging vehicles from the grid or rooftop solar, then reversing the flow of energy when needed. The system isolates each home from the grid during an outage, preventing any current from flowing into external power lines and endangering utility crews and nearby equipment.
“This project is demonstrating that bidirectional charging with CCS can work in occupied homes,” says Scott Samuelsen, founding director of the Advanced Power and Energy Program (APEP) at the University of California, Irvine, which is monitoring the two-year trial. “It’s a step toward vehicles that not only move people but also strengthen the energy system.”
For more than a decade, two-way charging has been available—but mostly restricted to Japan. Back in 2012 the Nissan’s LEAF-to-Home program proved the idea viable after the Tōhoku earthquake and tsunami, but that Nissan system relied on the CHAdeMO standard, little used outside of Japan. Most North American and European manufacturers chose CCS instead—a standard that, until recently, supported only one-way fast DC charging.
That distinction makes Menifee’s V2H-enabled neighborhood notable: it’s the first CCS-based V2H deployment in occupied homes, giving researchers real-world field data on a technology that’s been long trapped in pilot programs. The pairing of the Kia EV9 SUV with Wallbox’s commercially available Quasar 2 can deliver up to 12 kilowatts of power from the vehicle to the home.
It’s a step toward vehicles that not only move people, but also strengthen the energy system.”
–Scott Samuelsen, UC Irvine
Elsewhere, momentum towards commercial V2H has slowed. Ford’s F-150 Lightning supports home backup through Sunrun, but Sunrun equipment is not CCS-compatible. What’s more, Ford has announced a production pause for the pickup truck, which has delayed expansion. GM’s Ultium Home—a V2H system that works with the automaker’s Cadillac Lyriq, Cadillac Escalade IQ, Chevrolet Blazer, Chevrolet Equinox, Chevrolet Silverado, and GMC Sierra EVs— faces similar setbacks. Tesla’s PowerShare V2H feature is still stuck in a limited, early commercial rollout, with bidirectional compatibility restricted to the company’s Cybertruck. Menifee, by contrast, is producing operational data in real households.
When electric vehicles first hit the market, CCS was designed for one job: move power quickly from the grid to the car. The main goal was reliable, standardized, fast charging. That fact helps explain the difference between CCS public chargers, (many of which are rated for 350-kilowatts or more) and their CHAdeMO-based counterparts, which typically max out at 100 kW (but are capable of providing home backup or grid services).
Bidirectional operation wasn’t included in the original CCS standard for several reasons. Early automakers and utilities worried about safety risks, grid interference, and added hardware cost. So CCS’s original communication protocol linking EVs and charging stations—ISO 15118—didn’t even include an electronic handshake for power export. The 2022 update, ISO 15118-20, added secure two-way communication, enabling CCS vehicles to supply energy to buildings and the grid.
Wallbox’s Quasar 2 residential charger implements the update through an active-bridge converter circuit built with silicon-carbide transistors, achieving efficient bidirectional flow. Its 12-kW power rating can support typical critical loads in a house, such as heating and cooling, refrigeration, and networking, says Aleix Maixé Sas, a system electronics architect at Wallbox.
As the company’s name humbly suggests, Wallbox’s chargers look like plain old boxes—although they contain high-tech components.Wallbox
Each of the Menifee homes outfitted with a V2H system combines a rooftop solar array with a 13-kWh SunVault stationary battery from SunPower. During normal operation, solar energy powers daily household loads and charges the stationary battery. On abundantly sunny days, the solar panels can also top up the Kia EV9’s battery. When the grid fails—or when energy prices spike—the home isolates itself: Solar power and energy stored in the SunVault keep essential systems and appliances going, while the EV battery extends power if the outage persists.
This past summer, the UC Irvine researchers tracked how solar output, stationary storage, and vehicle power interacted under summer demand and wildfire-related grid stress. They found that “the vehicle adds a major resilience feature,” according to Samuelsen, who is the Menifee project manager. “It can relieve grid strain, increase renewable utilization, and lower costs by supplying power during peak-rate hours.”
Home builders and the makers of electric vehicle service equipment such as Wallbox are not the only entities reconsidering how to meet the engineering demands V2H introduces. Utilities, too, must make changes to accommodate bidirectional power flow. Interconnection procedures and energy pricing structures are among the factors that must be redesigned or reconsidered.
Analysts expect double-digit annual growth in bidirectional-charging system sales through the late 2020s as costs fall and standards mature. In regions facing wildfire- or storm-related outages and steep time-of-use pricing curves, projects like Menifee’s are showing a clear path towards the use of cars as huge and flexible energy reserves.
When EV batteries can supply energy for homes as easily as they do for propulsion, the boundary between transportation and energy will begin to disappear—and with it, old concepts regarding who’s an energy supplier and who’s a customer.
This article is crossposted from IEEE Spectrum’s careers newsletter. Sign up now to get insider tips, expert advice, and practical strategies, written in partnership with tech career development company Taro and delivered to your inbox for free!
At its core, engineering is an act of creation. This is why many of us chose to become engineers: We love to build things.
But especially if you have a private sector job, it’s easy to forget that passion to build as you climb up the corporate ladder. Somewhere between quarterly planning meetings and incident retrospectives, we often lose the joy of creation in our corporate jobs. Large companies require a level of bureaucracy and specialization that is often at odds with building something new.
That’s why I frequently recommend burned-out engineers to do something very simple: Start a side project. During 15 years working across various tech stacks and companies, this has been the most straightforward, underrated, and powerful way to regain my excitement at work.
Beyond rekindling a passion for creation, side projects have many other benefits. Side projects let us explore new technologies or problem spaces. We can leverage newer ideas that our companies may be hesitant to adopt. And you don’t need to get buy-in from a manager or explain the business justification. Start using a technology simply because you want to learn about it.
When you build something through a side project, your depth of understanding is far greater than just following a tutorial or reading about it. I can attribute many of my career opportunities to the projects I’ve built and published outside of my day job. Some of these projects, like my career growth platform Taro, even turn into companies!
We’ve entered the golden age for side projects because they’re so much more accessible. Compared to a decade ago, it’s significantly easier to research, build, and deploy your creation. Even compared to two years ago, you’re much less likely now to waste hours wrestling with some configuration rabbit hole. Just ask ChatGPT or Gemini for help!
The benefits of a personal project are real: passion, learning, career growth, and fun. And they’re easier than ever to create. Now’s the time to create your side project portfolio.
—Rahul
The quantum computing industry is growing, opening up new opportunities for engineers—and you don’t necessarily need a background in quantum physics to take these positions. So what skills do you need? See five key tips for breaking into the field from recruiters and researchers now working in quantum computing jobs.
Mini Thomas has built a highly successful career as an expert in power systems and smart grids—thanks in part, she says, to support from her family. Now a professor of electrical engineering in New Delhi, Thomas mentors women in the power industry, helping to expand the female leadership pipeline in India.
Have you received a notification from your bank or credit card company alerting you to suspicious activity on your account and requesting you confirm a purchase? You probably wondered how the bank suspected the charge wasn’t legitimate.
Credit card companies use a variety of methods to detect fraud, which is the most common type of identity theft and is on the rise, according to Experian, one of the major consumer credit information services.
Employer
Discover, in Raleigh, N.C.
Title
Data engineering manager
Member grade
Senior member
Alma mater
Christian College of Engineering and Technology, in Bhilai, India
To help prevent unauthorized transactions, IEEE Senior Member Pankaj Gupta is developing tools using data integration, artificial intelligence, machine learning, and real-time account monitoring. Gupta is a manager of data and analytics engineering for Discover, and he works from Raleigh, N.C.
“The innovations my fraud department has developed have helped my organization respond to threats faster and adapt more easily to future needs,” he says.
This year, he received Discover’s President’s Award, the company’s highest employee recognition. It is given to those who have achieved outstanding business results while demonstrating the company’s values.
Gupta also became an invited member of the Forbes Technology Council, a community of experienced leaders across industries, selected based on their professional achievements and leadership experience.
He says he enjoys his job but says he never meant to work in financial services, a field where he has built a nearly two-decade-long career.
As a youngster, he was curious about how things worked and would take apart gadgets his father brought home.
“My father worked at BSNL, a government telecom organization, and often took me to his office,” Gupta says. “There I would watch phones in operation and telecom operators connecting trunk calls. At home, we even had a few old, nonfunctional phones lying around.”
While at school, he enjoyed participating in science and math olympiads and exhibitions, he says.
Taking note of his curiosity and his problem-solving skills, his teachers encouraged him to study engineering. He was most interested in learning electrical engineering because of his interest in the power substations that he and his father checked out.
“Growing up in a small town [Dongargarh, which is famous for the Bamleshwari Temple, a popular Hindu pilgrimage site], I saw how technology could help solve problems and make life better for people,” he says.
He also became fascinated by the nearby steel factory’s massive machines, its chimneys, and the constant activity around them, he says.
“I noticed how everything seemed to work together in a coordinated way. Seeing such complex engineering in action at such a young age planted the seed of my interest in technology,” he says. “This exposure inspired me to pursue electrical engineering as a career.”
In 2002 he enrolled in the EE program at the Christian College of Engineering and Technology, in Bhilai, India, a 180-minute commute each way by train. He left at 5 a.m. and returned at 6:30 p.m. During his senior year, his family began struggling financially, he says, so his priority was to find a job immediately after graduating to support them.
In India, universities hold placement events on campus to recruit graduating students. Gupta says he was lucky enough to receive a job offer from the Indian IT company Satyam Computer Services, which is now defunct. He started there in 2006 after earning his bachelor’s degree in engineering and electrical engineering. Satyam assigned him to work on a software program for a financial services company.
That’s when he pivoted to software engineering.
Gupta says that although software development wasn’t his preferred career path, it enabled him to support himself and his parents.
He still has a soft spot for electrical engineering, he says, but he hasn’t changed fields or industries for the past 18 years.
His time at a variety of financial institutions has allowed him to travel the world to work in other countries including Germany and the United Kingdom, he says. The United States is the fourth country where he has worked.
“It’s been an exciting journey, learning about different work cultures and technologies,” he says.
Gupta left Satyam in 2011 to join Mphasis, an IT solutions company in Bangalore, India, as a senior software developer focused on extracting, transforming, and loading (ETL) data. After a year, he left for Wipro Technologies in Bengaluru. As a technical lead, he was assigned as a consultant for Capital One in Bangalore in an offshore development center. He led a team of 10 employees working on data integration projects including generic frameworks.
He moved to the United States in 2017 to work as an associate vice president at JPMorgan Chase in Jersey City, N.J. He helped create a world-class analytics platform and modernize the bank’s reporting systems. He also worked on systems that use a zero-trust security approach, which he describes as one whereby banks do not automatically trust any user or system. Instead, they verify every transaction.
“This greatly reduces the risk of fraud or unauthorized access,” he says.
He also developed scalable data partitioning techniques that organize and split large volumes of information into more manageable pieces.
“I believe that as AI advances, other innovations will evolve.”
“This allows the system to process data more quickly, handle growth without slowing down, and support real-time decision-making,” he says. “These innovations have helped my organization respond to threats faster.”
He joined Discover in 2019 and has worked his way up from principal data engineer for the data and analytics group to manager of data engineering. He developed AI-enhanced data pipelines to train models in making real-time, automated decisions.
AI and machine learning systems can prevent fraudulent transactions by collecting information about a customer’s typical financial habits over time, such as whether banking is done online or through an app, the time of day transactions occur, and the typical amount paid to creditors. The system is then trained to look for anomalies.
In simple terms, the system assigns a risk score to each transaction, usually on a scale based on patterns it has learned, Gupta says. If the score crosses a certain threshold, the bank might take preventive action. If, for instance, there is an unusual purchase in a location far from where the customer lives, the bank will send the customer a message, looking to verify whether the transaction is legitimate. If the customer does not recognize the purchase, the bank blocks the transaction.
Gupta joined IEEE in 2023 “to connect with a global network of technology professionals, and to stay updated in the latest advancements in engineering and computing,” he says. He was elevated to senior member later that year.
“Membership has helped me access world-class research through the IEEE Xplore Digital Library,” he says. “It also provided me an opportunity to attend conferences and share my expertise with the wider engineering community.”
His advice for young engineers is to stay curious and keep learning.
“Technology is changing very rapidly,” he says. “What is working right now might change in six months, so adaptability is your biggest strength.”
He predicts that AI agents will eventually take over repetitive tasks such as those related to automation, coding, and programming. Where engineers will be most needed, he says, is building AI models and training them.
“Engineers will find significant opportunities for growth in these areas,” he says. “I believe that as AI advances, other innovations will evolve.
“Focus on solving real problems, not just building solutions for their own sake.
“Build your professional network and seek mentors who can guide you through both technical and career challenges.”
When I traveled to Ellabell, Ga., in May to report on Hyundai Motor Group’s hyperefficient Metaplant—a US $12.6 billion boost to U.S.-based manufacturing of EVs and batteries—the company’s timing appeared solid. At this temple of leading-edge factory tech, Ioniq 5 and Ioniq 9 SUVs marched along surgically spotless assembly lines, giving the South Korean automaker a defensible bulwark against the Trump administration’s tariffs and onshoring fervor.
But dark clouds were already gathering. Consumer adoption of EVs had started slowing. The U.S. federal government’s $7,500 clean-car tax credit, which had helped hundreds of thousands of people make the leap to EVs, was being phased out.
Held securely on a yellow jig, a three-row Ioniq 9 SUV glides from station to station in the assembly hall. A view from below shows its generous, 110.3-kilowatt-hour battery pack, which, as in most EVs, sits below the floor of the car. The pack, which is shielded to prevent or limit damage in a collision, is part of an advanced 800-volt architecture for ultrafast DC charging. Christopher Payne/Esto
Near the Savannah-area factory, I drove a smartly designed Ioniq 9, a three-row SUV tailored to the United States’ plus-size tastes. I also saw a battery plant taking shape: a $4.3 billion joint venture between Hyundai and LG Energy Solution, on track to produce lithium-ion cells for Hyundai, Kia, and Genesis models in 2026. That facility is one of 11 low-roofed buildings that encompass 697,000 square meters (70 hectares), their pale green walls designed to blend into the Georgia countryside.
Backed by $2.1 billion in state subsidies, the Metaplant is the largest public development project in Georgia’s history. Covering 70 hectares, it is the centerpiece of Hyundai’s $12.6 billion total investment in the state, including the battery factory built with LG Energy Solution that ICE and other agents raided in September. Christopher Payne/Esto
That battery plant made headlines in September, when U.S. Immigration and Customs Enforcement (ICE) agents staged a workplace raid that led to more than 300 South Korean workers being detained and deported.
The episode highlighted the transnational cooperation—and tensions—inherent in importing a leading-edge manufacturing operation, a duality that might be familiar to anyone old enough to recall Japan’s game-changing entry into the U.S. automobile market in the 1970s and ’80s. The Metaplant is the largest publicly backed project in Georgia’s history. Its creation was accelerated by the Biden administration’s pro-EV policies, and it was also the centerpiece of Republican Gov. Brian Kemp’s bid to make his state “the electric mobility capital of the country.” Now, it was suddenly the latest flashpoint in an ongoing culture-and-trade war.
An automated guided vehicle (AGV) prepares to pick up a rack of windshields from an automated trailer unloader, for “just in time” delivery to an assembly line where Ioniq 5 EVs are being built. There is no human intervention from the time parts arrive at the Metaplant’s loading docks to their installation. Christopher Payne/Esto
Robots perform myriad tasks, yet human hands are still best for precision work. Jerry Roach, the Metaplant’s assembly manager, says, “I want my people doing craftsmanship. I want to pay people well for the things humans do well, and take away the stuff that’s tedious and boring.” Christopher Payne/Esto
As with other EV makers facing hurricane-force headwinds, including the U.S. rollback of pollution and fuel-economy rules, Hyundai has chosen to forge ahead with its long-laid plans. Company executives call the Metaplant North America’s most automated car factory and the most advanced full-scale factory among Hyundai Motor Co.’s 12 global manufacturing facilities. It rivals or surpasses Japan’s most advanced plants, such as the best operated by Toyota. Compared with the near-Dickensian Detroit auto factory that I toiled at in the 1980s, the stunning facility is a veritable MOMA: a modern museum of manufacturing art.
To have any chance of one-upping China, car factories elsewhere must become hyperefficient, which includes enlisting armies of AI-controlled robots—robots that can potentially work 24/7 and never ask for a raise or a lunch break.
The factory may eventually employ 8,500 people directly, and 7,000 satellite workers, for an annual capacity of 500,000 cars—more than Tesla’s Texas Gigafactory but less than Tesla’s Shanghai plant. This past summer, just 1,340 humans were sufficient to send a constant stream of two Ioniq models down these gleaming assembly lines. The “Meta Pros” working on those lines were earning on average $58,100 a year, which is 35 percent higher than the average in Bryan County, Ga.
Clearly the days of Ford’s River Rouge complex, which employed more than 100,000 in the 1930s, are gone. As in many new factories, you’ll see surprisingly few people beyond the assembly line itself. During my visit, I spotted less than two dozen in a cavernous welding hall, where 475 robots were piecing together car chassis in a whirling, metallic dance. A steel stamping plant was so quiet that no ear protection was required, even as robots stamped out roofs and other body panels, and then stowed them in overhead racks.
Outside, human workers parked their cars beneath solar roofs that generate up to 5 percent of the plant’s electricity. Meanwhile, a fleet of 21 hydrogen fuel-cell trucks, from the Hyundai-owned Xcient, carries parts from suppliers, emitting zero tailpipe emissions. The automaker’s goal is to obtain 100 percent of the Metaplant’s energy from renewables by 2030.
An Ioniq 9 body-in-white, the basic steel skeleton of an automobile, leaves the “main buck” section of the body build line. This line is where the vehicle’s floor and sides meet to form a recognizable car. The line adapts to changing production mixes to meet customer orders, with built-in flexibility to assemble future models.Christopher Payne/Esto
Sparks fly as welding robots piece together the Ioniq 9’s “body-in-white,” the industry term for the basic steel skeleton of a car, prior to the addition of subassemblies such as the suspension, power train, body trim, and interior. The Metaplant’s welding shop houses about 500 industrial robots.Christopher Payne/Esto
Robotic welders have revolutionized car manufacturing, joining the parts of an auto body with levels of speed, precision, and safety that humans can’t match. Such advantages reduce labor costs and scrapped materials. Hyundai is also now experimenting with humanoid robots to perform welding tasks.Christopher Payne/Esto
“Body-complete” robots mount front doors onto Ioniq 5s, using machine vision and laser-measurement systems to ensure an exact fit of movable panels on each body. The robots also install mounting bolts to exact torque specifications, all validated to ensure their work meets safety and quality standards.Christopher Payne/Esto
When those trucks roll into docks at the Metaplant, some of the factory’s 850 robots promptly unload their parts. About 300 automated guided vehicles, or AGVs, glide silently across the factory floor with no tracks required, trained to smartly stop for humans. An AGV rolls beneath a finished Hyundai, squeezes the wheels in its robotic arms, then swiftly hoists and ferries the car where it needs to go. A companion AGV further down the line executes the exact same moves. I’ve never seen so many robotic sleds like these, or a tag team move with more efficiency and grace. Within an AI-based procurement-and-logistics system, the AGVs allocate and deliver parts to workstations for “just in time” delivery, avoiding wasted time, space, and money as they stockpile components.
An automated guided vehicle ferries dashboards for the Hyundai Ioniq 9 SUV, including each dashboard’s pair of 30-centimeter display screens. AGVs are programmed to navigate the factory, using cameras and sensors to slow or stop to avoid collisions, and emit spoken warnings to human workers in their path.Christopher Payne/Esto
“They’re delivering the right parts to the right station at the right time, so you’re no longer relying on people to make those decisions,” says Jerry Roach, senior manager of general assembly at the Metaplant.
Roach prefers that his skilled humans focus on craftsmanship, doing jobs with tactile precision that only human hands and vision can accomplish. The idea is to free people from those elements of factory work that are physically taxing, unfulfilling, and, well, robotic, so workers can use their brains and take pride in their specialized skills.
Left: Adjustable-height carriers elevate an Ioniq 5 for easy access to the central fasteners and plugs that will position suspension components and the high-voltage battery, prior to the “marriage” between the upper and lower sections of the vehicle. Those carriers provide flexibility for automated functions and manual operations by the human workers at the plant (whom Hyundai calls Meta Pros). Right: On the final assembly line, an Ioniq 9’s “top hat”—including body panels—is married to the lower “skateboard” structure, which includes the electric motors, battery, and suspension. A finished car then undergoes various tests, including a water bath to check for leaks and a quick road test outdoors. Christopher Payne/Esto
Robots, Roach says, are best tasked with heavy lifting and repetitive tasks, or those that demand digitized speed and accuracy. One example is a “collaborative” robot, sophisticated enough to work safely in close proximity to people, despite its mammoth strength. For the first time at a Hyundai factory, such a robot is installing bulky, heavy doors on the assembly line—a notoriously tricky task to perform without scratching the glossy paint or damaging surrounding panels.
Hyundai is proud of its collaborative robots, including one that can precisely install a heavy door, a tricky task for humans to perform without damaging the panels. Those robots require advanced control systems so that they can work alongside human workers without needing to be fenced off or otherwise isolated.Christopher Payne/Esto
“Guess what? Robots do that perfectly, always putting the door in the exact same place,” Roach says. “So here, that technology makes sense.”
Man’s best friend, or its mechanical counterparts, stroll the factory floor: Spot, the robotic quadrupeds from Hyundai-owned Boston Dynamics, use camera vision, sensors, and what Boston Dynamics calls “athletic intelligence” to sniff out potential welding defects.
Spot, the robot dog designed by Hyundai-owned Boston Dynamics, inspects body welds on an Ioniq 5 for defects. Equipped with a sensor suite, the quadruped bot can recharge autonomously, dynamically work around fixed or moving obstacles, and get back on its feet if it falls. Christopher Payne/Esto
Those four-legged bots may soon have a biped master: Atlas, the humanoid robot, also from Boston Dynamics. The humanoid’s physical dexterity is uncanny, with a 360-degree swiveling head that allows it to walk forward and backward without turning its body. One look at these Atlases crawling, cartwheeling, or breakdancing during testing and you might reasonably conclude they’re a potential Terminator of jobs. Hyundai executives insist that’s not the case, even as they plan to put Atlases to work in their global factories. Boston Dynamics is training these robots to sense their environments and manipulate and move parts in complex sequences.
At this backup station, high-voltage battery fasteners can be installed in an Ioniq 5. The station ensures that the assembly line keeps running even if an automated production system requires servicing. Christopher Payne/Esto
From nearby Interstate 16, Georgia drivers can see freshly painted Ioniq 5s and 9s moving along a conveyor on a windowed bridge—an intentional glimpse of what’s happening inside. They can also see their tax dollars at work, after $2.1 billion in state subsidies. Hyundai is already building a second battery plant in Georgia, and a steel plant in Louisiana, part of an expanded pledge of $21 billion in U.S. investment through 2028.
After their frames are fully welded, Ioniq 5s move along a conveyor [in the background] to an environmentally friendly paint shop. From there, the cars will travel along an elevated bridge, visible from nearby Interstate 16 in Ellabell, Ga., toward final assembly.Christopher Payne/Esto
An Ioniq 5 arrives at its final inspection station. Immediately after, a human driver gets to drive the pristine car for the first time, on a test track just outside the factory. The first Ioniq 5 rolled off the Metaplant line on 3 October 2024, with the larger Ioniq 9 kicking off production in March 2025. Christopher Payne/Esto
In a suddenly inhospitable climate for EVs, there’s nothing automatic about building and selling the cars. But Hyundai and other automakers will keep trying. They don’t have any other choice.
It’s a little after 6:30 on a brisk July morning in a stone hut high in the Italian Alps. A gently hissing wood fire is leaking some warmth out of a brick oven. Gathered near it, around a big wooden table, some of Europe’s brightest young lepidopterists are doing what they do best: arguing in Spanish, Italian, and English about moths.
The Alte Pforzheimer Hütte, a stone house originally built in 1901, served as a base camp for the lepidopterists hunting rare moths in the Italian Alps.Luigi Avantaggiato
Scattered across the top of the table are dozens of moths in plastic specimen jars, the harvest of the previous night’s trapping. At one end of the table, Gioele Moro of the Czech Academy of Sciences is gently prying loose moths from the depths of a trap. At the other end, Laura Torrado-Blanco of the University of Oviedo’s entomological collection is paging through Lepidoptera guide books. She’s using the books to identify species—up here at 2,300 meters, there is no Internet connection.
A few of the scores of moths captured on a single night at a site in the Italian Alps are lined up on a bench in the stone hut. Researchers will identify the moths’ species and some of the insects will be sent on for tissue sampling and eventual genome sequencing. Luigi Avantaggiato
Looking up from a book, she notices me noticing the big butterfly tattoo on her left arm. “Chapman’s ringlet,” she tells me. “Erebia palarica,” she adds reflexively.
Pep Lancho Silva, a doctoral student at the Institute of Evolutionary Biology in Barcelona, extends a finger toward me with a spectacular creature on it: a large bone-white moth, with a black head and big black splotches on its wings. Torrado-Blanco is pretty sure it’s Arctia flavia, a species of tiger moth found only in rarefied air. If so, it’s precisely the kind of insect they came up here, to this chilly hut on the edge of a crystalline Alpine pond, to capture.
A yellow tiger moth, Arctia flavia, is among the catch at the stone hut, at an altitude of 2,300 meters.
At the crack of dawn in the stone hut, researchers [from left] Eric Toro Delgado, Laura Torrado-Blanco, Mónica Doblas-Bajo, and Gioele Moro (standing) unpack and examine the moths captured during the previous night.Luigi Avantaggiato
Lepidopterists have trapped, identified, and classified moths and butterflies for centuries. But this high-altitude confab is no Victorian perambulation. It’s a vital component of a sprawling, cutting-edge project that is pushing the boundaries of bioinformatics and the tools of modern genomics. These researchers are taking part in the first international field expedition of Project Psyche, whose goal is to sequence the genomes of all 11,000 species of moths and butterflies in Europe. Psyche is part of a larger effort, the Darwin Tree of Life project, which is itself a component of arguably the most ambitious science project of all time: the Earth BioGenome Project. Its goal is to sequence the genomes of all of Earth’s roughly 1.8 million organisms—every named species of animal, plant, fungus, and microbe that’s made up of cells that have a nucleus.
None of these hugely ambitious efforts would be conceivable without the enormous advances in genome sequencing and bioinformatics over the past couple of decades. The cost and speed of sequencing an individual genome have declined to the point where it’s now possible to batch process multiple genomes in a single day and for less than US $1,000 apiece. And the revolutions in biotech that have made such a feat possible are still gathering steam. Indeed, Earth BioGenome officials freely admit that their bold goal—to sequence those 1.8 million named species by 2035—won’t be possible without a hundredfold decrease in the time and cost of sequencing.
But the project’s success may ultimately hinge on functions other than sequencing. For example, after a creature’s genome is sequenced, the huge mass of raw genetic data—consisting of millions or billions of genetic building blocks called base pairs—must be annotated. That is, the tens of thousands of genes that make up the genome must be identified, located on chromosomes, and their functions or purpose described. And, of course, before an organism’s genome can be sequenced, its tissues must be sampled. To do that, researchers must locate the organism and, if it’s an animal, capture it. As I discovered with the Psyche team in the woods, valleys, and jagged peaks of South Tyrol, wrangling insects presents challenges that can defy logistics, technology, and even reason.
When I first heard about Project Psyche, the first thing I wondered was, Why Lepidoptera? I put the question to Charlotte Wright and Joana Meier at the hotel in Malles Venosta, Italy, that served as the headquarters for the Project Psyche expedition. They lead the project from its base at the Wellcome Sanger Institute in Cambridgeshire, England. The reasons, they tell me, span a range from pure science to completely commercial.
At the Hotel Tyrol in the Italian Alps, lepidopterist Charlotte Wright of the Wellcome Sanger Institute, a leader of Project Psyche, dissects the yellow tiger moth captured near the stone hut. Packed with liquid nitrogen, the tissue samples will subsequently be sent to the institute in England for genome sequencing.Luigi Avantaggiato
The earliest Lepidoptera appeared 250 million to 300 million years ago. By studying and comparing the genomes of different species, Wright explains, “we can find out how they have evolved and how they’ve diversified, as there have been different climatic shifts in Europe. And the genomes can help to tell us why it is that some groups of Lepidoptera have evolved into a greater number of species than others.”
Those genomes will also offer insights into some of the most intriguing questions of evolutionary biology. Consider: Most moths and butterflies have genomes with around 31 pairs of chromosomes, which are the threadlike strands in every cell’s nucleus, each of which is a molecule of DNA. Collectively, chromosomes make up a creature’s genome. But a tiny minority of the Lepidoptera order have enormous numbers of chromosomes. Exhibit A is the Atlas blue butterfly, which has an astonishing 229 pairs of chromosomes.
The Atlas blue is “a very good example of something that’s really fascinating, but we cannot understand it just by looking at one species,” says Meier. “What we really need is what Psyche will provide, which is replications”—thousands of Lepidoptera genomes. And, not incidentally, the ability to browse them easily. “Then we will find many lineages that have an unusually large number of chromosomes, and we can then start to ask, ‘What changes each time? What do they have in common? Do they have a repair gene that’s broken?’ ”
Some exceptional samples of Lepidoptera are preserved for entomological archives.Luigi Avantaggiato
And it’s not just theoreticians eagerly awaiting such genomic data. One practical aspect of these studies has to do with moths’ impact on agriculture. “There’s billions and billions of euros lost because agriculturally, some species do a lot of damage,” says Meier.
Adds Wright, “Pests are moving to new regions where previously they weren’t present and causing huge losses because the crops there haven’t been developed to be protected against these new species.” The reasons why some species succeed in a new area as climate changes, and are able to adapt and thrive, are also understandable only by studying many genomes—of the creatures that succeed, as well as the ones that don’t. “It’s kind of a dynamic situation, of monitoring these pests’ movements,” says Wright.
Shortly before sunset, Gioele Moro, of the Czech Academy of Sciences, sets up a moth trap on a mountain slope above the stone hut (the Alte Pforzheimer Hütte) in the Italian Alps. Luigi Avantaggiato
That, it turns out, takes a small army of grad students, researchers, and even citizen-scientists. Indeed, one of the goals of this expedition is to develop and refine best practices in collecting samples for genome sequencing and to train a cadre of young lepidopterists, who have varying levels of familiarity with the technologies of genome sequencing and annotation. On such techniques rests the success of not only Project Psyche, but also, ultimately, the Earth BioGenome Project.
It’s late in the afternoon of our first day in the high-altitude hut. Moro, of the Czech Academy of Sciences, is standing on a steeply raked mountainside in a dazzling sea of wildflowers—purple, yellow, lavender, crimson—that are gently swaying in the fading amber light. He’s wearing a black camp shirt, black cargo shorts, black socks, black hiking boots, and chunky retro eyewear, and he’s carrying a butterfly net (yep, it’s black). He’s still and silent, taking in nuances of light, vegetation, and wind that would affect a moth’s flight path through the area. Thinking like a moth, he visualizes the routes it would likely take through side valleys and ravines.
The objective is to figure out where to place three butterfly traps for the night. Setting the traps in different “microenvironments,” he explains, will likely yield a broader range of creatures. But there’s no formula for this. Capturing critters depends heavily on intuition arising from experience, perception, and judgment.
Genetics researcher Noé Dogbo, of the Institute of Research on Insect Biology in Tours, France, chases a butterfly during a hunting session in the Roja mountains near Curon Venosta, Bolzano, Italy. Luigi Avantaggiato
“Over there”—he points across the valley to the opposite slope. “It faces north. See? No flowers. That’s what I mean by different microenvironments.” We’re perched on the south-facing slope, about 80 meters above the valley bottom, on a trail about as wide as a toaster oven.
Hours later, after dodging cow patties the size of dinner plates and gaping holes leading to marmot burrows, the locations are chosen and the traps are set. There’s one on the south slope, one on the north, and one near the fast-flowing stream between them. As the sky darkens to a deep blue, we trudge back to the hut to stoke the fire and wait.
At the crack of dawn the next day, Moro is jubilant as he returns with the night’s haul. There are at least 150 moths, including the spectacular yellow tiger moth. The species that are needed for Project Psyche, as identified by Torrado-Blanco, are put in plastic specimen jars and will make their way down to the makeshift lab at the Hotel Tyrol. There, they’ll be photographed and then stunned and killed by exposure to dry ice, before being dissected. The head, thorax, and abdomen will be packed in separate plastic tubes for state-of-the-art DNA and RNA sequencing at the laboratories of the Wellcome Sanger Institute. The Wellcome Trust is the lead sponsor of both Project Psyche and the Darwin Tree of Life project.
Lepidopterist Joana Meier of the Wellcome Sanger Institute, a leader of Project Psyche, packs the abdomen of a moth into a vial for shipment from Italy to the institute in England. A bar code on the vial contains information about the sample and allows it to be tracked on its journey to the lab. Luigi Avantaggiato
The plastic tubes are packed in liquid-nitrogen-cooled shipping containers for the trip to Wellcome Sanger. DNA begins to break down almost immediately after death, especially in soft tissues. So the cryogenics are necessary to ensure that the samples arrive at Wellcome Sanger with as little degradation as possible.
Niklas Wahlberg of Lund University, in Sweden, is officially a “sampling hub leader” of Project Psyche. Unofficially, he’s one of the select few grizzled veterans here in Malles Venosta helping to mentor the young researchers, whose attendance is being funded through a European Union program called European Cooperation in Science and Technology.
Niklas Wahlberg, an evolutionary biologist at Lund University in Sweden, captures a moth in a plastic container at a trapping site along an Alpine trail above Malles Venosta, Italy.Luigi Avantaggiato
Wahlberg is an unabashed fan of moths. It’s not that he dislikes butterflies, mind you, it’s just that he’s a bit weary of them overshadowing moths in the public imagination. Butterflies are big, bright, and colorful, sure, but also delicate. They appeared much, much later than moths in evolutionary history. And they can’t even fly at night or in the rain. “Butterflies are just day-flying moths,” Wahlberg quips. “People think of them as different and special, but they’re not.”
In this new era of mass genome sequencing, they’re also arguably less important scientifically. To begin with, butterflies are just 10 percent of all known species of Lepidoptera—about 19,000 are butterflies while perhaps 180,000 or more are moths. Of the 11,000 European Lepidoptera species that are of interest to Project Psyche, only 560 of them are butterflies, by Wahlberg’s reckoning. And they’ve already collected two-thirds of them, he adds.
So the real challenge for Psyche is finding and identifying all those moths. Particularly the micromoths.
Micromoths have long vexed entomologists. The largest of them have wingspans about as wide as a U.S. dime, or a 2 euro cent coin; the smallest can fit on the head of a pin. As a group, they evolved not only much earlier than butterflies but also much earlier than all other moths (which are known as “macromoths”). There are a lot of micromoths—at least 62,000 species, by the current estimate. Among them are many pairs or other small groups of species that are so similar that not even the most experienced lepidopterists can tell them apart by eye.
Charlotte Wright of the Wellcome Sanger Institute collects a moth at a light trap on an Alpine trail above Malles Venosta, Italy.Luigi Avantaggiato
That’s going to be an enormous challenge for Project Psyche, Wahlberg notes. Fortunately, though, it’s a problem for which there is a technological solution: DNA barcoding.
Besides the DNA in the nuclei of every cell, there exists other genetic material, called mitochondrial DNA, outside of the nucleus. It’s relatively easy to access, and, crucially, there’s a mitochondrial gene, called CO1, that tends to vary markedly among species, even closely related ones. That makes this bit of genetic material invaluable for discriminating among related species. Researchers have built up several databases of these DNA barcodes that collectively contain millions of characteristic DNA sequences. “We have DNA barcodes for 99 percent of the Lepidoptera in Europe,” Wahlberg says. “And only about 5 percent of micromoth species have the same CO1 gene.”
DNA barcoding was invented in the early 2000s by Paul Hebert and colleagues at the University of Guelph, in Canada, and it has advanced greatly in recent years along with the DNA-sequencing technologies that underpin it. The technique starts with a minuscule sample of tissue; for example, in the makeshift lab at the hotel in Malles Venosta, researchers dissecting moths for sequencing also removed, for DNA barcoding, a leg of each moth whose species was not conclusively known.
Staff Scientist Silvia Pérez Lluch of the Centre for Genomic Regulation in Barcelona retrieves tissue samples for genome sequencing. To minimize degradation of the DNA in the samples, they are stored at -80 °C.Luigi Avantaggiato
Genetic material is isolated from that tissue, and then a CO1 gene is “amplified,” or replicated into many millions of copies, using a standard biotechnical technique called polymerase chain reaction. That material is sequenced using any one of the dozen or more types of sequencing machines available to researchers.
For barcoding purposes, typical DNA sequences of the CO1 gene run between 400 and 800 base pairs. But lately researchers have been developing techniques that use shorter or longer barcodes. The shorter codes, called mini-barcodes, have proven more effective in identifying a species even when the DNA samples are incomplete or damaged. A mini-barcode might have 100 to 250 base pairs. Conversely, “super-barcodes,” which can be many thousands of base pairs, are useful for differentiating among closely related species—exactly the challenge with many of the micromoths.
While the Psyche researchers honed the logistics and mechanics of sampling Lepidoptera, a different European Lepidoptera project was quietly making a technical advance that could resonate throughout the Earth BioGenome Project. Working together, Spanish and Andorran researchers affiliated with the Catalan Initiative for the Earth BioGenome Project sequenced the genome of the violet copper butterfly, Lycaena helle, a creature that was first studied in 1775. They described their efforts in a paper published by F1000Research.
This was no routine procedure. Typically when researchers map a genome, an organism is sampled and the DNA is sequenced. After sequencing, the mass of fragmented genetic data must be assembled into a complete genome sequence and then that complete sequence must be manually verified, in a process called curation, and then annotated. In annotation, the genome’s many genes are identified and, ideally, their functions described.
Ivo Gut, director of the Centro Nacional de Análisis Genómico in Barcelona, has high hopes for an emerging technique to identify the genes within a large mass of genetic data.Luigi Avantaggiato
Today, curation and annotation are time-consuming processes, regarded as major bottlenecks to the rapid progress that the Earth BioGenome Project desperately needs to reach its 2035 goal. Finding the thousands of genes within the huge mass of sequenced data is a mostly automated process now but it can involve some serious bioinformatic sleuthing. “You take your linear genome, your sequence, and you go and you say, ‘Ah, look here. There’s a gene that starts here,” says Ivo Gut, director of the Centro Nacional de Análisis Genómico (CNAG), in Barcelona. “ ‘And this is the structure of the gene.’ And then you can sort of figure out what that is. You look whether that gene is known, for example, in another species. And then you go to the next one, and so on. And just by these similarity searches, you can usually annotate almost 80 percent, or maybe 70 percent,” of what are known as coding genes in the genome. These coding genes encode the many proteins produced by cells, which serve vital functions in the organism.
Gut also notes that to perform annotations researchers are making increasing use of another genetic molecule, RNA, or ribonucleic acid. When a gene specifies, or “expresses,” a protein, RNA acts as the “messenger,” carrying the genetic code outside of the cell nucleus to the protein-making apparatus of the cell. Therefore RNA is extremely useful in figuring out where the protein-coding genes are in the genome. Different cells in the body express different proteins, but in every case that expression occurs because of a specific gene, and that gene can be identified conclusively from the RNA associated with it.
The breakthrough in the research by the Spanish and Andorran researchers was using a technique called long-read sequencing to sequence all of the RNA in their samples. While sequencing a genome, long-read machines handle much longer segments of DNA than traditional short-read systems. The greater length confers several advantages, including the ability to easily resolve repetitive sequences that can trip up short-read machines. [For more on long-read genome sequencing, see my recent article “The Quest to Sequence the Genomes of Everything, in IEEE Spectrum.”] The researchers came from four Barcelona organizations—CNAG, the Centre for Genomic Regulation (CRG), the Institute of Evolutionary Biology at Pompeu Fabra University, and the University of Barcelona—and from Andorra Research and Innovation, in Sant Julià de Lòria.
The genome of the female violet copper butterfly, which inhabits a huge swath of territory stretching from the Pyrenees to Siberia, consists of 25 pairs of chromosomes with a total of 547,306,268 base pairs. By using long-read sequencing of the RNA in the sample, the researchers were able to identify 20,122 protein-coding genes and 4,264 noncoding genes. In contrast to protein-coding genes, noncoding genes are harder to identify from one species to the next and they are also very difficult to predict by computational means. Many noncoding genes serve important regulatory, protective, or other functions within a cell. Yet at least 30 percent of all annotated Lepidopteran genomes produced so far lack annotations of noncoding genes, and those that include them generally count relatively few, says Roderic Guigó Serra, who leads the Bioinformatics and Genomics program at the CRG.
“Long-read RNA sequencing may be the only way to precisely locate them in genome sequences,” he says. With long-read RNA sequencing, “we get better information on where the genes are and a more precise definition of the boundaries of the genes, and also we see genes that had not been seen before,” Serra declares.
At the Guigò Lab of the Centre for Genomic Regulation in Barcelona, a technician loads a sample into a genome sequencing machine. Luigi Avantaggiato
His group is now applying the long-read RNA sequencing technique to a host of other species—including humans. They’re doing this through Gencode, an international consortium that aims to produce improved, “reference” annotations for the human and mouse genomes. Twenty-five years after the first draft sequence of the human genome, it turns out that there are still gaps in it—particularly regarding the noncoding genes. Recently, using long-read RNA sequencing, the Gencode team shocked biologists by identifying 18,000 previously unknown noncoding human genes. “These genes have been essentially ignored for almost 25 years, underscoring the power of the long-read RNA sequencing technology,” says Serra.
Researchers are counting on such advances to help power them in their grand quest of sequencing and annotating the world’s organisms. And within that quest, Project Psyche is off to an encouraging start. With nearly 3,000 of Europe’s 11,000 Lepidopteran species sampled and more than 1,000 of those sequenced, Lepidoptera are now the most widely sequenced order of organisms. Still, that leaves perhaps 170,000 other members of the order elsewhere in the world to be sampled and sequenced.
It’s a mammoth task. As they grapple with it, its practitioners can take inspiration from the novelist and lepidopterist Vladimir Nabokov. “My loathings are simple,” he wrote in 1973. “Stupidity, oppression, crime, cruelty, soft music. My pleasures are the most intense known to man: writing and butterfly hunting.”
As an electrical engineering student in the 1980s and ‘90s, Carlotta Berry had two experiences that helped shape her future as an educator.
First, while she studied robots, she wasn’t allowed to interact with them. “The robots were too expensive, so the undergrads did not get to touch them,” Berry recalls. “I said to myself, I’m going to teach engineering someday, but in a way that the students will get to touch and program the robot.”
This led Berry to work toward overcoming the economic exclusivity of robotics. But her second formative undergrad experience involved a different type of exclusion: Berry was one of only a few engineering students who were female or Black. “It sometimes could be a lonely experience,” says Berry. “Representation does matter.”
Now, Berry is a professor in the electrical and computer engineering department at Rose-Hulman Institute of Technology, where her students learn about human-robot interactions and mobile robotics by using actual robots.
Berry works on her first open-source modular 3D-printed robot, the LilyBot, with Rose-Hulman engineering students Murari Srinivasan (left) and Josiah McGee (right). Bryan Cantwell/Rose-Hulman Institute of Technology
She also works to support people of color in engineering. Almost three decades after she graduated, Berry realized little progress had been made when she heard Black women grad students describe feeling isolated and marginalized during an online engineering conference in 2020. “This was exactly how I felt 30 years ago,” says Berry, noting that today only about 8 percent of electronics engineers are women and about 5 percent are Black. “It was time for something to change.”
As a child in Nashville, Berry excelled at school—especially math—and thought she’d become a math teacher. But in high school, a mentor suggested that Berry consider engineering, given her strong grades in both math and science. “I didn’t really know what an engineer was,” she recalls. “I didn’t know anyone who was an engineer.” After learning about the profession at a library, Berry decided to study both engineering and math in college. In 1993, Berry earned a bachelor’s in electrical engineering at the Georgia Institute of Technology as part of a dual degree program with Spelman College, where she earned a bachelor’s in mathematics in 1992.
After her bachelor’s degrees, Berry worked as a control engineer for Ford Motor Company, where she programmed assembly-line industrial robots, but she found herself yearning to answer her true calling as an educator. So, she returned to academia and got a master’s in electrical engineering and control systems at Wayne State University in 1996. Saddled with student loan debt, however, Berry then accepted a position as a control engineer for Detroit Edison. “I really enjoyed the work but once again realized I was not doing what I was meant to do,” she says.
After a year at Detroit Edison, she left in pursuit of her Ph.D. in electrical and computer engineering, which she earned at Vanderbilt University in 2003. As a grad student, Berry taught at a technical school—and at last found herself on the right career path: “I always wanted to be an educator,” she says.
Berry traces her community-outreach work to two more pivotal moments in her career: In 2018, she became a full professor at Rose-Hulman, and in 2020, she became an endowed chair in the Electrical and Computer Engineering department. Berry says her tenure and position at Rose-Hulman enabled her to pursue work that brings her research, teaching, and service interests together.
Berry hopes to support women of color in STEM through public events. Here, she sits with students Liz Francois and Janae Gillus, both members of Rose-Hulman’s chapter of the National Society of Black Engineers.Griffin Museum of Science and Industry
“As a full professor, I don’t have to worry that someone might consider the [outreach] work I do not as important as my technical robotics work,” she says. “When I provide education for students and for the community, that’s also part of my research and service.” For Berry, research and service are not separate but intertwined subject areas: Her research involves designing open-source, low-cost mobile robots to promote more inclusive robotics education.
Since 2020, Berry has helped transform how electrical and computer engineering is taught and perceived. She has been teaching hands-on, interactive robotics not only to her students at Rose-Hulman but also to kids and adults across the nation. Berry has been taking her robots, as she says, “to the streets.”
Berry demonstrates and discusses her open-source, 3D-printed wheeled robots at schools, libraries, museums, and other community venues. Her audiences range from kids just a few years old to adult educators who learn about robotics from Berry so they can teach the subject to their own students. To spread the word about robotics and STEM, Berry also has become active on social media, overcoming her innate introversion because, she explains, “visibility matters.”
With any audience, Berry is always “very approachable and very engaging,” says Nicki Manion, a program manager for Rose-Hulman’s educational outreach who collaborates with Berry on professional development workshops for teachers.
“I have to go where people are,” Berry says. “I get robots in front of people who are historically marginalized and would normally not have access to these technologies.”
This past summer, for example, Berry shared her robots with children from about three to 10 years old at all of the dozens of branches of the Indianapolis Public Library. To understand the three main pillars of robotics—sense, plan, act—the kids learned how the robots use a sonar, microphone, and speaker in order to see, hear, and talk. Notably, at the end of each presentation, the kids got to play and interact with the robots.
Last year, as part of an IEEE Education Society Initiative, Berry brought her robots to the streets globally. After grad students in countries such as Costa Rica, Niger, and Uganda received parts in the mail, Berry showed them the basics of building and programming robots.
Berry hasn’t set out on her pedagogic journey all on her own, she says. In 2020, she cofounded Black in Engineering and Black in Robotics—part of the Black in X network comprising more than 80 organizations that support the work of Black professionals in STEM. For Berry, it’s no coincidence that Black in X emerged early in the pandemic. “There were a lot of bad things about the pandemic, but because we were all home and on social media, we were able to connect and find each other and form these organizations that, five years later, are still going,” she says.
Her professional turning point toward more community-oriented service has led to several accolades, she says: “That was when I started to earn these awards I had never been considered for before.” In 2023, the IEEE Robotics and Automation Society awarded Berry the prestigious Undergraduate Teaching Award for her contributions to multidisciplinary robotics education and leadership in diversifying STEM. She has also been recognized by the Society of Women Engineers and AnitaB.org.
Children’s books like the series Berry wrote help get kids interested in STEM.Rebellion LIT
On top of her outreach and community work, Berry finds time to write children’s books—work that also has its roots in the pandemic. During that time, Berry woke up from a dream and remembered only the title of a children’s book she knew she had to write: There’s a Robot in My Closet. The book spawned a series, which features kid protagonists learning how to program robots and developing their problem-solving skills. (Berry also writes STEM-centered romance novels for adults under the pseudonym Carlotta Ardell. The heroine of her book Elevated Inferno, Berry says, struggles with the expectation to flawlessly juggle work and life—an expectation that falls more heavily on women, she finds.)
While balancing her many personal and professional interests, Berry says, she maintains a clear-eyed pursuit of her professional mission: helping people of diverse backgrounds “see themselves as not just consumers of technology but creators of technology.”
I was a little disappointed by China’s World Humanoid Robot Games.1 As fun as real-life Rock ‘Em Sock ‘Em Robots is, what people really care about is robots doing their chores. This is why robot laundry folding videos are so popular: we didn’t know how to do that even a few years ago. And it is certainly something that people want! But as this article so nicely articulates, basic laundry folding is in a sweet spot given the techniques we have now. It might feel like if our AI techniques can fold laundry, maybe they can do anything—but that isn’t true, and we’re going to have to invent new techniques to be really general purpose and useful.
With that in mind I am issuing a challenge to roboticists: Here are my Humanoid Olympic events. Each event will require us to push the state of the art and unlock new capabilities in robotic manipulation. I will update my Substack as folks achieve these milestones, and will mail actual real-life medals to the winners.
In order to talk about why each of these challenges pushes the state of the art in robotic manipulation, let’s first talk about what’s working now. What I’m seeing working is learning from demonstration. Generally, folks are using puppeteering interfaces. Most common seems to be two copies of the robot so that a human can grab and move one of them while the other follows, or a virtual reality headset with controllers for hand tracking. They then record some 10-30 second activity hundreds of times over. Fromm that data, a neural network is trained to mimic those examples. This technique has unlocked tasks that have steps that are somewhat chaotic (like pulling a corner of a towel to get it to lay flat) or have a high state space (like how a towel can be bunched up in myriad different ways).
Thinking about this method of training robots to do things, it should be clear what some of the limitations are. Each of these has exceptions, but together they form a general trend.
Now, on to the events!
Things like doors are tricky because of the asymmetric forces: you need to grasp and twist the handle or knob quite hard, but if you pull hard outside of the arc of the door you tend to slip your grasp. Also, moving through a door requires whole body manipulation, which is more than I’ve seen from anyone yet.
Benjie Holson
I think this is very close to state of the art (or maybe it has happened and I didn’t see it). I expect this medal to be claimed by December.
Benjie Holson
Adding self-closing makes this significantly more challenging because of the force involved, though the lever handle is arguably easier (I just don’t see many round-knob self-closing doors).6
Benjie Holson
The boss fight of doors.7 You need to either use a second limb to block the door from re-closing, or go through the door fast enough to use dynamics.
We’re just getting started on laundry.
Benjie Holson
This is probably doable using the techniques we have now, but it’s a longer horizon task and might require some tricky two-handed actions to pull the shirt through to right-side-out.8
Benjie Holson
I think both the hand-insertion and the action of pinching the inside of the sock are interesting new challenges.
Benjie Holson
The size medium shirt starts unbuttoned with one sleeve inside-out. It must end up on the hanger correctly with the sleeve fixed and at least one button buttoned. I think this one is 3-10 years out, both because buttons are really hard and because getting a strong, dexterous hand small enough to fit into a sleeve is going to be hard.
Humans are creatures of technology and, as useful as our hands are, we mostly use them to hold and manipulate tools. This challenge is about building the strength and dexterity to use basic tools.
Benjie Holson
The window cleaning fluid bottle is super forgiving in terms of how you grasp it, but you do need to independently articulate a finger, and the finger has to be pretty strong to get fluid to spray out.9
Benjie Holson
The challenge here is to pick up a knife and then adjust the grasp to be strong and stable enough to scoop and spread the peanut butter. Humans use a ‘strong tool grasp’ for all kinds of activities, but it’s very challenging for robot grippers.10
Benjie Holson
A keyring with at least 2 keys and a keychain is dropped into the robot’s waiting palm/gripper. Without putting the keys down, get the correct key aligned and inserted and turned in a lock. This requires very challenging in-hand manipulation, along with high precision forceful interaction.
We humans do all kinds of in-hand manipulation using the structure of our hands to manipulate things that we are holding.
Benjie Holson
Requires dexterity and precision, but not very much force.
Benjie Holson
When I use a dog-bag I have to do a slide-between-the-fingertips action to separate the opening of the bag which is a tricky forceful interaction as well as a motion that I’m not even sure most robot hands are capable of. Also tricky is tearing off a single bag rather than pulling a big long spool out of the holder, if you choose to use one.11
Benjie Holson
Done without external tools. This is super tricky: high force yet high precision fingertip actions.
If you sit down and write out what you might want a robot to do for you, a lot of tasks end up being kind of wet. Robots usually don’t like being wet, but we’ll have to change that if we want to have them clean for us. And wet things can be difficult to grasp and use.
Benjie Holson
Mildly damp, but with exciting risk of getting the whole hand in the water if you aren’t careful. Probably requires at least splash resistant hands (or a whole bunch of spares).
Benjie Holson
This one naturally follows after the sandwich one. Water everywhere. Seems like an important skill to have after a few hours collecting training data on the dog-poop task.
Benjie Holson
Water, soap, grease, and an unpleasant task no one wants to do.
To be eligible to win, a general purpose manipulator robot running autonomously must demonstrate successful task completion in a real-time video with no cuts. You are allowed a maximum of 10x the time I took to do each task (a 4 second task can take your robot up to 40 seconds). I reserve the right to be arbitrary in deciding if things aren’t following in the spirit of the challenge. First robot to achieve this wins the prize!
To claim your medallion email [email protected] with an address for me to ship it to. If you give me a photo of your robot wearing a medal I will be tickled pink. I will also accept future challengers that are at least 25% faster than the current winner. Some medals have already been claimed; you can see the winning videos here. Good luck and may the odds be ever in your favor.
Thanks to Jeff Bingham for advice, fact checking and cool robot videos. And thanks to my patient wife for spending an hour filming me doing silly things in a silly costume.
1 As far as I can tell, kickboxing was just the Unitree mini-humanoid robot, and everyone had the same code running, so… I guess it won?
2 TRI has some pretty cool stuff with force control using a big training rig.
3 Tesla’s Optimus has 22 degrees of freedom using cable drives (cause you can’t fit those motors in a hand). In 2008 I worked on this robot which also had 22 degrees of freedom and controlling it was crazy hard (as was keeping all the cables correctly tensioned). The other hand was a big two-finger gripper which I ended up using for most teleop tasks.
4 Meta has been working with some in-finger vision systems which seem cool.
5 This is likely more a teleoperation precision limitation than a model limitation. Here is a video of Generalist Robotics doing sub-cm precision tasks. I love that hockey sticks have become the traditional “mess with a robot” tool even for ridiculous things like this.
6 Yes, I did wear this at my workplace in order to get this video. You’re welcome.
7 I have programmed (not trained) a general purpose mobile manipulator to pass through a self-close pull door, but it took over 4 minutes (disqualified for taking too long) and required a special doorstop. Also the video isn’t public (also disqualified). Also it’s really tacky to put up a competition and award yourself gold before it even starts.
8 T-shirt starts fully inside-out in a wad. Finishes tolerably folded, right-side out.
9 You must spray 3 good spritzes on the window, and wipe them up with paper towels so there are no ugly streaks. Paper towels start on the paper-towel roll, not pre-torn and pre-wadded.
10 Peanut butter jar starts and ends closed. Sandwich should be cut in half. (Triangle or rectangular cuts are both acceptable, though your three-year-old might disagree).
11 Mock poo allowed. Bag starts on the roll but can be in a standard dog-bag holder, held by the robot.
This post originally appeared on General Robots, Benjie Holson’s Substack about making a general purpose robot company.
