Our research goal is to develop novel computational approaches to designing and controlling mobile robots, especially for tasks in challenging environments such as forests and underwater. To do this, we identify environment structures that can assist mobile robots in localization, navigation, and other sub-tasks and we intentionally design the robot hardware and software to leverage this environment structure. The most famous examples of this approach are robot vacuums – the early models did zero mapping or localization and relied on stochastic coverage guarantees to guarantee they will eventually cover an entire floor just by turning randomly when a bump sensor is triggered.
By minimizing the resources used by each robot, we can make multi-robot systems more scalable (both to larger numbers and smaller robots). Our research group also looks at distributed algorithms for large multi-agent collectives, with a special focus on maintaining formal guarantees on collective behavior.
Part of achieving real-world robustness and safety involves the principled design of interfaces and specifications where humans are in the loop, so our work also involves human-computer/robot interaction (HCI/HRI) and programming language / API development.