Obstacle avoidance framework based on reach sets

Abstract

This paper reports preliminary investigations concerning the development of a LIDAR based sense and avoid (SAA) system with a low computational footprint for small Unmanned Air Vehicles (UAVs). The focus is on the integration with nominal flight control systems and on computational feasibility. The proposed system decomposes the SAA problem into the following components detection, space assessment, escape trajectory estimation and avoidance execution. The control logic is encoded with the help of a hybrid automaton. The properties of the system are studied with the help of approximations to time slices of the UAV reach set.