An Egocubemap Based Algorithm for Quadrotors Obstacle Avoidance Using a Single Depth Camera

International audienceA fast obstacle avoidance algorithm is a necessary condition to enable safe flights of Unmanned Aerial Vehicles (UAVs) eventually at high-speed. Large UAVs usually have a lot of sensors and available computational resources which allow complex algorithms to run fast enough to navigate safely. On the contrary, small UAVs gather many difficulties, like computation and sensors limitations, forcing algorithms to retain only a few keys points of their environment. This paper proposes an obstacle avoidance algorithm for quadrotor using a single depth camera. Taking advantage of the possibilities offered by embedded GPUs, a cubic world representation centered on the robot-called Egocubemap-is used while the whole obstacle detection and avoidance algorithm is light enough to run at 10 Hz on-board. Numerical and experimental validations are provided