Research
Robotic systems have the potential to enhance our way of life and solve some of the world’s most difficult problems—but in doing so they encounter situations unanticipated by the robot designer and software developer. We are interested in bridging this gap by framing artificial curiosity and experimentation as real-time computational problems solvable through online optimization and control.
Our research focuses on the analysis of physical dependencies in algorithmic design, the relationship between sensing capabilities and potential learning outcomes, and minimizing the computational and data requirements for robots to achieve complex tasks and behaviors. Our work spans a wide range of problems in robotics, optimal control, and learning including sample-efficient learning and exploration, coverage, manipulation, locomotion, and multi-agent coordination.