Skip to Main Content
This paper considers the task of computing (sub-) optimal continuous and discrete control trajectories for hybrid systems with a probabilistic discrete transition structure. The discrete inputs are used to block undesired transitions and to realize evolutions that are goal-attaining with high probability. An approach is proposed which combines the computation of discrete shortest-paths for given probability-levels with embedded continuous optimal control problems to fix the continuous controls. The result is a control strategy which leads to a system evolution that maximizes a weighted sum of performance and probability of success. The approach is illustrated for an automated transportation scenario.