Correct and efficient UAV missions based on temporal planning and in-flight hybrid simulations

Abstract

Controller synthesis has been successfully applied in UAV applications, to construct a mission plan that is guaranteed to be correct with respect to a user-provided specification. Albeit being correct, these plans may not be optimal in the vehicle´s trajectory, battery consumption, or other criteria which the user may consider relevant. A possibility would be to apply a quantitative synthesis approach where the target is to compute efficient plans before the mission, at a higher cost of complexity and potential limitations in the optimization goals to achieve. As an alternative, in this paper we propose doing the plan optimization in-flight. For this, we use available tools that synthesize controllers with multiple controllable choices and later select among these choices in-flight using hybrid simulations ranking them according to the optimization objective. We present the advantages of our approach and validate them using software-in-the-loop simulation with typical UAV mission scenarios.