A viability approach for fast recursive feasible finite horizon path planning of autonomous RC cars

We consider a viability based approach to guarantee recursive feasibility of a finite horizon path planner. The path planner is formulated as a hybrid system for which a difference inclusion reformulation is derived by exploiting the special structure of the problem. Based on this approximation, the viability kernel, which characterizes all safe states and the corresponding safe controls, can be calculated. Using the set of safe controls the computation time of the on-line path planning can be reduced, by only generating viable trajectories. Finally, a condition characterizing the unsafe set in case of on-line obstacle avoidance is derived.