Kinodynamic motion planning for high-dimensional physical systems

This thesis presents a kinodynamic motion planner, Kinodynamic Motion Planning by Interior-Exterior Cell Exploration (KPIECE), specifically designed for systems with complex dynamics, where physics-based simulation is necessary. A multiple-level grid-based discretization is used to estimate the coverage of the state space. The coverage estimates help the planner detect the less explored areas of the state space. The planner also keeps track of the boundary of the explored region of the state space and focuses exploration on the less covered parts of this boundary. Extensive experiments show KPIECE provides computational gain over state-of-the-art methods and allows solving some harder, previously unsolvable problems. A shared memory parallel implementation is presented as well. This implementation provides better speedup than an embarrassingly parallel implementation by taking advantage of the evolving multi-core technology. Acknowledgements First of all, I would like to thank my advisor, Dr. Lydia Kavraki, for her guidance and help with this work, and the Physical and Biological Computing Group for their valuable comments and suggestions. I would like to thank Mark Yim and Jonathan Kruse for their help in defining the CKBot ODE model. Many thanks go to Marius S ¸ucan for designing most of the images this document includes, and to Derek Ruths for naming the algorithm (KPIECE).