My research focuses on algorithms to enable decision theory —famous for its potential but notorious for its computational hardness— become practical software tools in robotics. Such tools will enable robots to design their own strategies for accomplishing the tasks we want them to do, despite various modelling errors and uncertainty in the system and its operating environment. Currently, I focus on algorithms to generate optimal motion strategies in the presence of uncertainty (including uncertainty due to modelling errors, sensing and actuation errors, and unpredictability of the operating environment), and on their applications in robotics and animal locomotion.

Active Projects

(warning: a bit outdated, will be updated soon)

Practical POMDP Software

Optimal Inspection Planning

Animal Locomotion In-Silico: A Tool to Study Collision Avoidance Strategies of Flying Animals

Past Projects