Optimal motion planning of wheeled mobile robots

The thesis presents some methods useful for the optimal planning and control for the motion of autonomous wheeled vehicles. In particular, the exposed techniques may be applied to the wide class of flat systems. Results can be summarized as an hybrid feedforward/feedback control scheme, whose purpose is to guarantee a robust and highly performing control. High performances are reached out with the planning of time-optimal and continuous velocity profiles and geometrically continuous paths, that lead to a continuous steering input signal. This means that a smooth and optimal motion of the wheeled vehicle can be attained and, in such a way, the vehicle autonomous navigation can perform agile and event-driven maneuvers. Robustness is achieved by means of iterative trajectory replanning procedures, which guarantee the tracking of the planned trajectory in the presence of noise. It has been proved the existence, for the proposed trajectory planning methods, of closed-form bounds on the tracking error. Simulation and experimental results obtained during this research point out that the presented methods may be well suited for a real-time implementation provided that some of the required optimizations are done off-line. Indeed, optimal velocity profiles and paths can be generated in real-time using fast local optimization routines based on look-up tables built with off-line optimization.