Why the future needs robots with a human touch
Roboticists for decades have focused their attention on making machines that can improve human life. However, achieving the payoff on that idea — with robots performing complex, diverse tasks in real environments with humans present — is still a challenging feat at hand.
This could be due to how researchers have historically investigated the part about “how to make the machines work on their own” (robotic autonomy) with much more enthusiasm than having explored how the field of robotics can be used to “improve human life” (collaborative robotics). Roboticist Monroe Kennedy III is preparing to change that.
To Kennedy, the challenges of moving from autonomous robotic platforms to collaborative robotic technologies bring a considerable delay in progress. He believes that in the next few years robots should be designed to address challenges in our communities (for example, providing aid in assisted living facilities with direct impact on aging populations); deliver dexterity and precision to enable in-house production of products and devices in companies; and help improve workplace safety by providing essential robotic assistance (completion of service tasks).
Robots as teammates
A major goal is to realize the pioneering and enterprising vision of enabling robots to assist in the tasks they are trained and needed to do. Pursuing this ambitious vision, Kennedy is moving forward from the essential robotic autonomy questions of how to perform basic functions toward focusing on tools and algorithms that enable robots to serve as teammates with humans, and to provide strong support and partnership in the work and home environments.
To advance the field of robotics, Kennedy needs to finish building the technology that will serve as the catalyst for this important evolution in robotics. And he’s close! This spring, his team at Stanford’s Assistive Robotics and Manipulation Laboratory (ARMLab) will finish the design for a soft robotic finger almost as sensitive and stabilizing as the digits on a human hand. Kennedy says achieving this robotic sense of touch will be an important contribution to the field of collaborative robotics. “What’s keeping these robots from doing useful things for us?” Kennedy asks. “If we had to pick one thing, it would be the sense of touch.”
In March, Kennedy’s team debuted a new tactile sensor (DenseTact) that enables a robot’s fingers to accurately model surfaces in high resolution, giving the robot a better sense of the object that it’s holding and improving its dexterity. His team’s 3D sensor uses a fisheye camera to image the surface of anything the robot touches, then relies on deep neural networks to estimate in real time how the robot’s fingers will be impacted by contact with the object. This improves how the robot adjusts its grip to handle many different objects with ease. With this design, he’s given robots a more accurate sense of touch at a fraction of the cost of existing sensors.
Once the design for this sophisticated sense of touch has been optimized for both high-resolution force and shape sensing, robots could perform a variety of complex tasks, from setting a glass of water on a nightstand for a high-risk patient in a health care setting to smartly assembling devices to reduce a company’s manufacturing costs. Next his team plans to take that technology and discover ways to scale it, so researchers everywhere can plug that sense of touch into any robotic design, and any robot can learn to manipulate any object. It could be the breakthrough needed to focus the entire field on beneficial applications of decades of robotics research – and point experts at new problems. “If that's achievable now, roboticists could ask questions that they couldn’t ask before,” Kennedy says.
Diversity of ideas
To support this research over the next five years, Kennedy received a National Science Foundation CAREER Award in 2022. With the award, Kennedy not only continues his research, but also expands the diversity of ideas pouring into his lab. For example, this year he created a new Stanford course, Collaborative Robotics, which introduces students to new ways of thinking about robotic assistants. At the end of the course, students in their final projects consider how to push the field forward, and Kennedy expects these student projects will inspire new research directions for ARMLab, with opportunities for students to contribute.
The funds will also support Monroe’s efforts to increase diversity in the research lab, and help bring into focus organizations like Black in Robotics, a separate nonprofit organization Kennedy co-founded in response to the murder of George Floyd in 2020 to advance diversity, equity and inclusion in the field of robotics. “It’s a very important point, the role that diversity plays in robotics,” Kennedy says, explaining how prioritizing diversity fits into his vision for more rapidly advancing robotics to better serve communities. “You need people from different backgrounds, different cultures, people who think differently, to see a problem from all different types of perspectives, in order to generate solutions that are as far-reaching as possible.”
He says his passion for inventing began in high school, when his scientist parents introduced him to physics and electrical engineering. While pursuing his bachelor’s in mechanical engineering at the University of Maryland,Baltimore County, he co-founded an inventors’ club, then took an internship in roboticist Vijay Kumar’s lab. While there, he saw how life-changing robotic technology could be, so he continued his studies in mechanical engineering and applied mechanics (MEAM) at the University of Pennsylvania, where he received his MS in robotics and PhD in MEAM. Then, after six months as a researcher at the Massachusetts Institute of Technology’s Lincoln Lab, Kennedy joined Stanford University as an assistant professor of mechanical engineering in 2019.
Inspiring the next generation of roboticists
At Stanford, he saw an opportunity to directly inspire the next generation of roboticists and, working with them, edge collaborative robotics forward to achieve the maximum community impact faster than expected. At ARMLab, the goal is not only to theorize but also to actually develop collaborative robots that can improve human life. Over the next five years, he plans to take the calibrated model from his soft robotic finger and apply the technology to other parts of the robot’s body, creating a sensitive skin that could cover a large surface area and help robots better perform in unique environments.
More generally, his priority is to create more equitable access to more advanced assistive technology and collaborative robotics to meet the huge demand that already exists today, which he says will only grow. As the number of aging adults in the United States is projected to hit nearly 90 million by 2050, Kennedy thinks it’s time we break the paradigm where we think of robots as a substitute for a pair of human hands, and instead start thinking of robots as helping hands that can fill the caregiver gap and help extend quality of life. “People think it’s robots versus people,” Kennedy says. “I think it’s better to think of robots and people.”