|
|
|
|
|
ABSTRACT For a variety of applications, multi-robot approaches offer advantages over single robots in robustness, efficiency, and even feasibility. At the distributed and reactive end of the multi-robot system spectrum, I will present mathematical modeling methodologies developed to predict and optimize a robotic swarm's performance for several tasks. The models deliver qualitatively and quantitatively correct predictions several orders of magnitude more quickly than an embodied simulator can. These methods are a useful tool for optimizing and generalizing the swarm behavior of these highly stochastic, asynchronous, nonlinear systems, often outperforming intuitive reasoning. As tasks and behaviors become more complex and completeness becomes a concern, the swarm modeling assumptions no longer apply. For a group of problems focused on multi-robot inspection planning, I will present an approach for abstracting each problem into two components: a graph representing the particular instance of the inspection task and a graph problem whose solution Represents a complete plan for inspection. Inspection plans are computed using a new constructive heuristic algorithm and simulations illustrate its performance and characteristics. SPEAKER BIO Kjerstin Williams is a Ph.D. candidate in Electrical Engineering at the California Institute of Technology, where she previously received her B.S. (2000) and M.S. (2002) degrees. She is currently working in the Burdick lab's multi-robot cooperation research group at Caltech, focusing on path planning for inspection tasks. Kjerstin Williams |
|
Back to CRES |