Potential fields-aided motion planning for quadcopters in three-dimensional dynamic environments

With the increasing use of unmanned aerial vehicles (UAVs), the safe operation and navigation of a UAV need to be guaranteed, and this requires a collision avoidance (CA) mechanism for UAVs. The artificial potential field (APF), a widely used CA approach, has some issues like local minima and infeasible trajectory. This paper proposes a novel approach to overcome those drawbacks by combining motion primitives (MP) and the APF. In fact, the MP generates a locally optimal and dynamically feasible trajectory for the given time duration. When the collision checker detects the risk of collision at sample points extracted from the planned trajectory, the best route among re-planned safe path candidates is selected using the APF. It is shown that the proposed approach provides smooth and feasible trajectories without any collision in three different scenarios.