Motion planning and series expansions for underactuated mechanical systems


Francesco Bullo
Coordinated Science Lab & General Engineering
University of Illinois at Urbana-Champaign

Advances in nonlinear control of Lagrangian systems are leading to a rigorous framework with applications to stabilization and motion planning problems. Autonomous vehicles, robotic manipulators and locomotion devices all share the same Lagrangian dynamics that can be exploited in control problems.

This talk focuses on motion algorithms for stabilization and control of underactuated autonomous vehicles. The main result are some planning schemes amenable to on-line implementation. Local motion primitives are designed to accomplish basic tasks such as changing and maintaining velocity. These primitives are then combined to specify low velocity maneuvers, such as point to point re-configuration and stabilization. Theoretical tools include the study of series expansions for the solution of time-varying differential equations, local nonlinear controllability, and motion planning for systems with drift.