A Symbolic Approach for Multi-target Dynamic Reach-avoid Problem

This paper develops a symbolic technique for the control and path planning of autonomous vehicles in adversarial environments. We consider the objective of the autonomous vehicle as visiting a set of targets sequentially, while avoiding a dynamic adversarial (defending) vehicle. To address this multi-target dynamic reach-avoid path planning problem, this paper proposes a novel efficient hybrid, correct-by-design, symbolic solution. For this purpose, the dynamic interactions between the vehicles and their logical behaviors are captured in the form of General Reactivity (1), and then a hybrid control system is constructed to generate winning trajectories, which satisfy the objective of the attacking autonomous vehicle. The proposed technique is verified through several simulations and the results show the effectiveness of the developed algorithm.