Computer Science Colloquium
Friday, September 30
2:35pm in Wege
“Walking softly: How compliance can improve locomotion for legged robots”
Traditional robots are rigid and operate in well defined, predictable environments. As we have begun to build for and deploy robots in the real world, roboticists have typically sped up the planning and control methods used in traditional robotics so that they can recover from mistakes faster rather than developing methods that are inherently robust to uncertainty. In this talk, Sonia will show how robots can use compliance to improve their ability to walk and run in the real world. She will use her PhD research on legged locomotion in deserts to motivate the need for new, inherently robust control methods, and then discuss how inherently compliant materials — namely, knitted fabric — could be used to build the robots of the future. Could we use knitting machines to “3D print” robots with arbitrary morphologies and compliance properties for different tasks?
Sonia Roberts is a postdoctoral researcher in the Electrical and Computer Engineering department at Northeastern University, where she works with Professor Kris Dorsey to develop soft, reconfigurable sensors with applications to robotics. She received her PhD from the University of Pennsylvania in 2021, and her dissertation work with Prof. Dan Koditschek improved the locomotion capabilities and energetic efficiency of legged robots on sand. Sonia’s undergraduate education in Cognitive Science gave her a long-standing interest in finding solutions to locomotion problems that do not require an agent to build complicated representations of the world. Sonia is a 2022 Robotics, Science & Systems Pioneer, a 2020 Rising Star in EECS, and a recipient of the Complex Scene Perception NSF IGERT Fellowship at Penn. Currently, Sonia is developing soft, knitted sensors that could be used in fully knitted robots.
Megamenu Social