Audrey Chen ’24 lives by the philosophy that “a lot of opportunities only present themselves if you ask for them.” This approach has served her well, from becoming a NASA intern at 15 to running MIT’s autonomous boat team Arcturus to entering a leadership position at 3D printing technology company Formlabs right out of undergrad.

Growing up in Los Angeles, Chen showed a strong aptitude and passion for engineering at a young age and skipped several grades in math. In her first year of high school, she saw a posting about the JPL Space Academy at NASA’s Jet Propulsion Lab. Though the program was for juniors and seniors, she inquired if they would make an exception for her and they agreed. By her junior year she was helping run the program as deputy.

But Chen didn’t stop there: She had dreams of interning at NASA. She asked her mentor and became a drone air traffic control researcher at NASA at 15. “I was not old enough to drive,” Chen says. “High school would end, the bell would ring, and I would put on my backpack and I would run down the street to JPL. Can you imagine you’re the security guard at the gate of the Jet Propulsion Laboratory and a kid shows up for work?”

Chen worked on the Orbiting Arid Subsurfaces and Ice Sheet Sounder (OASIS) project, whose goal is to find and examine freshwater aquifers and ice sheets. “It was very early in the mission, so I was doing system and objective definition,” Chen says.

Next stop: MIT

After graduating high school, Chen ventured across the country to explore her eclectic interests at MIT. When she wasn’t fulfilling the requirements for her mechanical engineering degree, she could be found leather crafting, glass blowing, or table welding in one of MIT’s makerspaces, documenting MIT student life with her camera (garnering the acclimation “The Eyes of MIT” by MIT Admissions), working as a researcher sampling deep-sea sediment, or notably, running the award-winning autonomous boat team Arcturus.

“Arcturus has been the highlight of my MIT career,” Chen says. She founded the team at MIT Sea Grant in 2022 along with a group of equally impassioned students who elected Chen as captain.

“I didn’t have any background in marine autonomy, so we pushed very hard to institute trainings and have lots of workshops so that they would feel comfortable coming in and contributing as soon as possible,” she recalls. Seeking additional funding and support, the team found a home at the MIT Edgerton Center.

Launching Arcturus

“Whenever I think about how Arcturus started and how it somehow still continues, I think it’s a miracle,” Chen says. “Our very first year, there were five of us at the Roboboat competition, and if any individual one of us had not decided to join the team, we either would not have a boat, we would not have electronics, we would not have code to run the boat, or we wouldn’t have funding to run the team.”

Chen’s first year as captain was a tremendous amount of work because the team was so small. In addition to managing the team and assuring they met their goals on time, Chen also acted as the team’s business lead, treasurer, media lead, and photographer. “I was juggling a lot of things. Since then, those roles have further split amongst more people within the team,” she says.

Recruiting isn’t easy for an autonomous boat team, as many students don’t get marine robotics experience in high school. To keep their recruitment pool wide, Chen didn’t expect students to have background in autonomy or in marine systems. “Creating an environment that’s welcoming and friendly and supportive of people’s learning is crucial, because otherwise you won’t have a team. We’ve really pushed hard to recruit from a large body of people. We make sure to emphasize that we’re open to all majors, all years. As an industry, marine robotics, like most engineering, is very male-dominated. We work hard to recruit people of all genders and ethnicities.”

With Chen’s skillful recruiting, Arcturus increased from five to 74 members in 2024. Arcturus flourished under Chen’s leadership, winning First Place Design Overall at the Roboboat competition in 2023.

The challenges with autonomous boats

Chen was drawn to autonomous boats because the field is so full of potential. “You leave a robot on land and you turn it off, it doesn’t move by itself, versus you put it in a body of water and you don’t do anything, then it still moves because of the currents. It needs to be constantly taking in that input and trying to localize where it is,” Chen says.

Chen sees a lot of potential in the marine biotics industry to gather crucial data about our environment. “Autonomy in the marine space is not as well researched as land autonomy is. There’s immense potential for marine autonomy to benefit the world. You think about mapping ocean topology or looking for endangered species or habitat protection or surveying bleached coral reefs. As a vehicle, you have more flexibility to move around versus a buoy. That gives you the ability to take water and sediment samples across a wider spread of area. And by making it autonomous, you eliminate high labor costs, so the price per sample for a researcher would go down. These are different ways in which autonomy has potential to benefit the research sphere, but also, more broadly, the world.”

Chen graduated early this past February and passed Arcturus on to captains and rising juniors Ami Shi and Karen Guo. “They’re rock stars. The team is in good hands,” Chen says.

Becoming a project manager at Formlabs

Chen graduated a semester early and accepted a project manager position at Formlabs. She brings many lessons from MIT to her work. “The biggest thing that I’ve learned is that I don’t need to know everything. Part of being successful is knowing what you don’t know. So I’m always aware that in every Arcturus meeting, and probably every technical meeting that I’ll be in at Formlabs, that I will not be the smartest person in the room. And that’s fine. I don’t need to be the smartest person ever because that’s not my job. My job is to bring these projects together and know enough about all the systems to integrate them.”

Chen is thrilled to stay near MIT after graduation, allowing her the opportunity to visit her friends and continue mentoring Arcturus. Upon announcing her new job, she remarked, “To my friends at MIT, I’ll be just down the street, so you won’t be able to get rid of me that easily!”

Japan’s Mission to Become a Global AI Powerhouse

Language Models Take Center Stage

Japan is on a mission to become a global AI powerhouse, and it’s starting with some impressive advances in AI-driven language models. Japanese technology experts are developing advanced models that grasp the unique nuances of the Japanese language and culture—essential for industries such as healthcare, finance, and manufacturing – where precision is key.

Partnerships and Innovation Hubs

But this effort isn’t Japan’s alone. Consulting giants like Accenture, Deloitte, EY Japan, FPT, Kyndryl, and TCS Japan are partnering with NVIDIA to create AI innovation hubs across the country. The centers are using NVIDIA’s AI software and specialized Japanese language models to build tailored AI solutions, helping industries boost productivity in a digital workforce. The goal? To get Japanese companies fully on board with enterprise and physical AI.

Omniverse Platform: A Game-Changer for Industries

One standout technology supporting the drive is NVIDIA’s Omniverse platform. With Omniverse, Japanese companies can create digital twins—virtual replicas of real-world assets—and test complex AI systems safely before implementing them. This is a game-changer for industries such as manufacturing and robotics, allowing businesses to fine-tune processes without the risk of real-world trial and error.

Addressing Challenges and Bridging the Gap

Japan faces a shrinking workforce presence as its population ages. With its strengths in robotics and automation, Japan is well-positioned to use AI solutions to bridge the gap. In fact, Japan’s government recently shared its vision of becoming "the world’s most AI-friendly country," underscoring the perceived role AI will play in the nation’s future.

Growing AI Market and Infrastructure

Japan’s AI market hit $5.9 billion in value this year; a 31.2% growth rate according to IDC. New AI-focused consulting centers in Tokyo and Kansai give Japanese businesses hands-on access to NVIDIA’s latest technologies, equipping them to solve social challenges and aid economic growth.

Cloud Providers and Infrastructure

Top cloud providers like SoftBank, GMO Internet Group, KDDI, Highreso, Rutilea, and SAKURA Internet are also involved, working with NVIDIA to build AI infrastructure. Backed by Japan’s Ministry of Economy, Trade and Industry, they’re establishing AI data centers across Japan to accelerate growth in robotics, automotive, healthcare, and telecoms.

Conclusion

Japan is taking big strides toward establishing itself as a leader in the AI-powered industrial revolution. With its focus on advanced language models, partnerships with consulting giants, and cutting-edge technologies like Omniverse, Japan is well on its way to achieving its goal of becoming the world’s most AI-friendly country.

Frequently Asked Questions

Q: What is Japan’s vision for AI in the nation’s future?
A: Japan’s government aims to become "the world’s most AI-friendly country," with AI playing a key role in addressing social challenges and aiding economic growth.

Q: How is Japan’s AI market growing?
A: Japan’s AI market hit $5.9 billion in value this year, with a growth rate of 31.2% according to IDC.

Q: What role will cloud providers play in Japan’s AI development?
A: Cloud providers like SoftBank, GMO Internet Group, and others are working with NVIDIA to build AI infrastructure, establishing AI data centers across Japan to accelerate growth in various industries.

Since its launch in 2022, the MIT Morningside Academy for Design (MAD) has supported MIT graduate students with a fellowship, allowing recipients to pursue design research and projects while creating community. Pulling from different corners of design, they explore solutions in fields such as sustainability, health, architecture, urban planning, engineering, and social justice. 

On May 1, MAD announced the 2024 cohort of Design Fellows at the MIT Museum.

Play video

Meet the MIT MAD 2024 Design Fellows
Video: MIT Morningside Academy for Design

Sofia Chiappero, MCP student in the Department of Urban Studies and Planning and MITdesignX affiliate: Chiappero is working around the intersection of community development and technology, aiming to address the challenges faced by underserved communities at risk of displacement in Latin America. Through a blend of social science and digital inclusion, she seeks to design a new approach to researching human interactions and replicating them in virtual settings, with the ultimate goal of preserving the identity of these communities and giving them visibility for resilient growth.

Clemence Couteau, MBA candidate in the MIT Sloan School of Management: Couteau is tackling the rise of postpartum depression among U.S. mothers by aiming to develop a digital solution empowering at-risk pregnant women to improve mental health outcomes. This involves a self-directed therapy chatbot in a mobile app, based on the “ROSE” protocol.

Mateo Fernandez, MArch student in the Department of Architecture: Fernandez explores how to depart from the current construction industry, designing alternatives such as growing buildings with biomaterials, and deploying advanced 3D printing technologies for building.

Charlotte Folinus, PhD candidate in the Department of Mechanical Engineering: Folinus creates new methods for designing soft robots, using these tools to design soft robots for gentle interactions, uncertain environments, and long mechanical lifetimes. “I am really excited to be surrounded by people who can do things I cannot. That’s when I’m the best version of myself. I think that’s the community I’ll find here,” she says.

Alexander Htet Kyaw, master’s student in the Department of Architecture and the Department of Electrical Engineering and Computer Science and MITdesignX affiliate: Htet Kyaw’s current research utilizes robotic assembly, multimodal interaction, and generative AI to challenge conventional manufacturing and fabrication practices. He is working on an AI-driven workflow that translates design intent into tangible objects through robotic assembly.

Dení López PhD candidate in the Department of Urban Studies and Planning: As a Design Fellow, López uses design research to evaluate and extend the scope of Bicheeche Diidxa’, a long-standing Participatory Action Research initiative for disaster resilience focused on five Zapotec communities along the Los Perros River in Oaxaca, Mexico.

Caitlin Morris, PhD candidate in media arts and sciences: Morris’s research explores the role of multisensory influences on cognition and learning, and seeks to find and build the bridges between digital and computational interfaces and hands-on, community-centered learning and teaching practices.

Maxine Perroni-Scharf, PhD candidate in the Department of Electrical Engineering and Computer Science: Perroni-Scharf is currently working on developing techniques that enable the discovery and design of extremal metamaterials — 3D printed materials that exhibit extreme properties arising not from their chemical composition, but rather from their structure. These can be applied to a variety of tasks, from battery design to accessibility.

Lyle Regenwetter, PhD candidate in the Department of Mechanical Engineering: Regenwetter develops methods to incorporate design requirements, such as safety constraints and performance objectives, into the training process of generative AI models.

Zane Schemmer, PhD candidate in the Department of Civil and Environmental Engineering: Schemmer’s research aims to minimize the carbon footprint of the built environment by designing efficient structures that consider the availability of local materials.

For people with paralysis or amputation, neuroprosthetic systems that artificially stimulate muscle contraction with electrical current can help them regain limb function. However, despite many years of research, this type of prosthesis is not widely used because it leads to rapid muscle fatigue and poor control.

MIT researchers have developed a new approach that they hope could someday offer better muscle control with less fatigue. Instead of using electricity to stimulate muscles, they used light. In a study in mice, the researchers showed that this optogenetic technique offers more precise muscle control, along with a dramatic decrease in fatigue.

“It turns out that by using light, through optogenetics, one can control muscle more naturally. In terms of clinical application, this type of interface could have very broad utility,” says Hugh Herr, a professor of media arts and sciences, co-director of the K. Lisa Yang Center for Bionics at MIT, and an associate member of MIT’s McGovern Institute for Brain Research.

Optogenetics is a method based on genetically engineering cells to express light-sensitive proteins, which allows researchers to control activity of those cells by exposing them to light. This approach is currently not feasible in humans, but Herr, MIT graduate student Guillermo Herrera-Arcos, and their colleagues at the K. Lisa Yang Center for Bionics are now working on ways to deliver light-sensitive proteins safely and effectively into human tissue.

Herr is the senior author of the study, which appears today in Science Robotics. Herrera-Arcos is the lead author of the paper.

Optogenetic control

For decades, researchers have been exploring the use of functional electrical stimulation (FES) to control muscles in the body. This method involves implanting electrodes that stimulate nerve fibers, causing a muscle to contract. However, this stimulation tends to activate the entire muscle at once, which is not the way that the human body naturally controls muscle contraction.

“Humans have this incredible control fidelity that is achieved by a natural recruitment of the muscle, where small motor units, then moderate-sized, then large motor units are recruited, in that order, as signal strength is increased,” Herr says. “With FES, when you artificially blast the muscle with electricity, the largest units are recruited first. So, as you increase signal, you get no force at the beginning, and then suddenly you get too much force.”

This large force not only makes it harder to achieve fine muscle control, it also wears out the muscle quickly, within five or 10 minutes.

The MIT team wanted to see if they could replace that entire interface with something different. Instead of electrodes, they decided to try controlling muscle contraction using optical molecular machines via optogenetics.

Using mice as an animal model, the researchers compared the amount of muscle force they could generate using the traditional FES approach with forces generated by their optogenetic method. For the optogenetic studies, they used mice that had already been genetically engineered to express a light-sensitive protein called channelrhodopsin-2. They implanted a small light source near the tibial nerve, which controls muscles of the lower leg.

The researchers measured muscle force as they gradually increased the amount of light stimulation, and found that, unlike FES stimulation, optogenetic control produced a steady, gradual increase in contraction of the muscle.

“As we change the optical stimulation that we deliver to the nerve, we can proportionally, in an almost linear way, control the force of the muscle. This is similar to how the signals from our brain control our muscles. Because of this, it becomes easier to control the muscle compared with electrical stimulation,” Herrera-Arcos says.

Fatigue resistance

Using data from those experiments, the researchers created a mathematical model of optogenetic muscle control. This model relates the amount of light going into the system to the output of the muscle (how much force is generated).

This mathematical model allowed the researchers to design a closed-loop controller. In this type of system, the controller delivers a stimulatory signal, and after the muscle contracts, a sensor can detect how much force the muscle is exerting. This information is sent back to the controller, which calculates if, and how much, the light stimulation needs to be adjusted to reach the desired force.

Using this type of control, the researchers found that muscles could be stimulated for more than an hour before fatiguing, while muscles became fatigued after only 15 minutes using FES stimulation.

One hurdle the researchers are now working to overcome is how to safely deliver light-sensitive proteins into human tissue. Several years ago, Herr’s lab reported that in rats, these proteins can trigger an immune response that inactivates the proteins and could also lead to muscle atrophy and cell death.

“A key objective of the K. Lisa Yang Center for Bionics is to solve that problem,” Herr says. “A multipronged effort is underway to design new light-sensitive proteins, and strategies to deliver them, without triggering an immune response.”

As additional steps toward reaching human patients, Herr’s lab is also working on new sensors that can be used to measure muscle force and length, as well as new ways to implant the light source. If successful, the researchers hope their strategy could benefit people who have experienced strokes, limb amputation, and spinal cord injuries, as well as others who have impaired ability to control their limbs.

“This could lead to a minimally invasive strategy that would change the game in terms of clinical care for persons suffering from limb pathology,” Herr says.

The research was funded by the K. Lisa Yang Center for Bionics at MIT.