ECE Department Seminar: Planning, Decision Making & Learning for Robotic Information Gathering
March 20, 2013
Dr. Geoffrey A. Hollinger,
Viterbi School of Engineering
University of Southern California
Time: 3:00 PM -- 4:00 PM, Mar. 20th (Wednesday), 2013
Location: Babbio Center, Room 319
Typically when robots are tasked with gathering information (e.g., in urban search and rescue, environmental monitoring, and aerial surveillance scenarios), human operators must oversee almost every aspect of the operation to ensure completion of the task. Strict human oversight not only makes such deployments expensive and time consuming but also makes some tasks impossible due to the requirement for heavy cognitive loads or superhuman reaction times. These limitations can be mitigated by making the robotic information gatherers autonomous, reducing deployment cost and opening up new domains (e.g., underwater monitoring and space exploration).
However, the problem of optimizing robot motion plans to maximize information is extremely difficult due to the partial observability of the environment and the exponential growth of the planning space in both the length of the mission and the number of robots. With optimal solvers, it may take hours or even days to plan the actions of a small team of robots. On the other hand, heuristic solvers are computationally efficiently but may perform arbitrarily badly. The tradeoff between optimality and efficiency motivates the development of scalable robot planning algorithms with guarantees that perform near-optimally in practice.
In this talk, I will show how a general framework that unifies information theoretic optimization and physical motion planning makes autonomous information gathering tractable. I will leverage techniques from submodular optimization, adaptive decision making, and machine learning to provide scalable solutions with performance guarantees in a diverse set of applications such as underwater inspection, urban target search, marine monitoring, and sensing for sustainable energy. The techniques discussed here make it possible for autonomous robots to “go where no one has gone before,” allowing for information gathering in environments previously outside the grasp of human investigation.
Geoffrey A. Hollinger is a Postdoctoral Research Associate in the Viterbi School of Engineering at the University of Southern California. His current research interests are in adaptive information gathering, distributed coordination, and learning for autonomous robotic systems. His past research includes multi-robot search at Carnegie Mellon University, personal robotics at Intel Research Pittsburgh, active estimation at the University of Pennsylvania's GRASP Laboratory, and miniature inspection robots for the Space Shuttle at NASA's Marshall Space Flight Center. He has served as a guest editor for the Autonomous Robots journal and on program committees for the IEEE International Conference on Robotics and Automation (ICRA), the Robotics: Science and Systems Conference (RSS), and the International Joint Conference on Artificial Intelligence (IJCAI). He received his Ph.D. (2010) and M.S. (2007) in Robotics from Carnegie Mellon University and his B.S. in General Engineering along with his B.A. in Philosophy from Swarthmore College (2005).