My goal is to enable autonomous robots to work productively in our world, from carrying out our daily chores to performing dangerous tasks such as search and rescue missions. To operate effectively and efficiently, autonomous robots need to be able to sense, reason, and act. The ability to move reliably is critical for the success of the whole operation. Without this ability, the robot will not be able to accomplish the given tasks, not to mention work productively. My research focuses on computational representations and algorithms to generate reliable motion strategies for autonomous robots. I have been using random sampling in conjunction with efficient exploitation of domain structure to develop efficient motion planners.
During my Ph.D., I focused on path planning for robots with many DOFs. Path planning concerns with finding collision-free paths for robots to move from a given initial to a given goal configurations is an environment populated by static obstacles. After my Ph.D., I expanded the scope of my research to motion planning in uncertain and dynamic environment. In particular, I focus on enabling Partially Observable Markov Decision Processes (POMDPs), a general and principally mathematical framework for planning with uncertainty and partial observability, to be practical for real-world robotics problems.
