Autonomy is a critical requirement for robots in a variety of application domains ranging from search and rescue to planetary exploration. Even though recent advances in computing, actuation and energy storage technologies promise to yield platforms that are power autonomous with onboard computation, achieving true behavioral autonomy remains a formidable challenge.

One of my long-term research goals is to build a reasoning framework that is capable of accurately capturing dynamic properties of a system, yielding effective abstractions to physical behavior to enable their safe and robust composition. The distinguising feature of this approach is its emphasis on abstractions that are informed by the physical system rather than being motivated primarily by practical implementation issues.

This is an ambitious goal and it must draw upon tools from control theory, dynamical systems and computer science. I will update this page as these ideas become more concrete.