Learning and Control for Human-Centered Autonomy
Abstract: Robots have been successfully deployed in controlled, robot-centered environments. The next frontier lies in developing intelligent robots capable of operating in open-world, human-centered environments, where they can assist and serve humans, generating broad societal benefits. Achieving this vision requires a paradigm shift in control design—from focusing solely on robots to modeling and controlling the complex mix-autonomy system. It remains challenging to create robots that can safely operate around humans and effectively serve every individual’s utility. In this talk, I will present my research to addressing these challenges, with an emphasis on application in autonomous driving. First, I will introduce our approaches for synthesizing controllers catering to every individual user’s requirement, leveraging the synergies of human data, reinforcement learning, and model predictive control. Second, I will briefly summarize my efforts on improving the robustness of data-driven human traffic models by improving their representations of human interactions. I will conclude by outlining next steps toward the widespread adoption of robots in open-world, human-centered environments, including transportation systems and beyond.
Bio: Chen Tang is a Postdoctoral Fellow in Computer Science at UT Austin. Prior to that, he was a Postdoctoral scholar in Mechanical Engineering at UC Berkeley. He received his Ph.D. in Mechanical Engineering from UC Berkeley in 2022 and his bachelor’s degree in mechanical engineering from HKUST in 2016. He received the ASME DSCD Rising Star Awards in 2022 and was selected as an RSS Pioneer (in Robotics) in 2023. His research interest lies at the interaction of control, robotics, and learning, with applications in autonomous driving and social robot navigation.
Zoom link: https://wpi.zoom.us/j/93413349160