Expansion rate based collision avoidance for Unmanned Aerial Vehicles

Collision avoidance is a critical problem for Unmanned Aerial Vehicles (UAVs) for its wide application. We proposed a vision based collision avoidance approach for UAVs, based on potential field method. By exploring human navigation model, the potential field is designed over the UAV´s heading using the relative heading to the goal and to obstacles, the distance to the goal, and the expansion rate of obstacles. This method is independent of any information of obstacles, and controls the UAVs heading directly away from the obstacle and towards the goal. The simulation results show that the paths are smooth and much more optimal and robust compared with the model with angular width.