The research focus of this project is the study of two sets of problems at the frontiers of modern motion planning. These are planning with kinodynamic constraints and planning with high dimensional reparameterizable robots with complex constraints imposed by the energetics of an underlying physical system. Both sets of problems define high dimensional state spaces. At its core, the project will develop general solution frameworks based on randomized sampling algorithms for the exploration of these high dimensional state spaces together with performance measures and guidelines on how to choose relevant parameters. To obtain a healthy balance between studying fundamental issues and specialized applications, representative probems have been identified. For kinodynamic planning, the orbital rendezvous problem will be studied, where a robot space vehicle in some orbit docks with a space station in a different orbit. For planning with physical constraints, knot tying will be investigated. The proposed project is rooted in motion planning but incorporates concepts from control theory, radomized algorithms, computational geometry, geometric modeling and graphics, and complexity theory.
The broader impact part of the project is implemented through (a) plans for technology transfer to industrial, government and biomedical research laboratories, (b) curriculum development at Rice University involving undergraduate robotics and theory courses, (c) training, mentoring and involvement in research activities of undergraduate, graduate and postdoctoral students, (d) mentoring of women undergraduate students in computer science, and (e) participation in an NSF funded program of Rice University and the Houston Independent School District whose goal is to attract high school girls to engineering fields.
