Formation Path Planning using Model Predictive Control

Abstract

The use of Unmanned Aerial Vehicle (UAV) formations has gained great interest over recent years due to their potential applications in both military and civilian uses, including surveillance, mapping and search-and-rescue missions. UAV formations refer to multiple cooperative UAVs which move together to achieve a collective goal, while staying close and maintaining a specified geometric shape. Researchers worldwide work to find effective and robust solutions for the formation flight problem. The main challenges include finding a safe path for the formation to fulfill the given task and controlling the UAVs to keep a specified relative position to each other during the motion of the formation. Here, Model Predictive Control (MPC) emerged as a promising approach for path planning for UAVs. The optimal path is generated by repeatedly solving an optimal control problem, taking the UAVs’ dynamics and physical limitations into account, as well as other constraints, e. g., for obstacle and collision avoidance. We propose MPC for the path planning of UAV formations. By directly considering the formation-keeping objective, the formation control is integrated into the path planning process. In this work, we present two different methods that extend an existing MPC-based path planning algorithm designed for area coverage problems in 2D environments. Both methods use a centralized MPC approach, in which the paths of each UAV in the formation are computed simultaneously. The first method is based on the leader-follower approach, in which one UAV, denoted as the leader, covers the search area and the other UAVs follow to maintain a desired relative position to the leader. In the second method, all UAVs are equal agents flying in a polygonal formation and working cooperatively to cover the search area. The performance of the proposed methods is tested in different simulated scenarios with a formation of three UAVs. The UAVs show the desired behavior of flying in formation, while avoiding collisions with obstacles and other UAVs. During obstacle avoidance maneuvers the formation shape is temporarily disrupted, in order for the UAVs to avoid a collision with an obstacle, however, is restored after the maneuver. By using a centralized MPC approach, the overall behavior of the formation is optimized, because the given objectives, including keeping the formation shape, as well as obstacle and collision avoidance constraints are considered simultaneously.