Global and Local Fusion Method for Path Planning and Formation Control of Multiple Unmanned Systems

This paper studies a formation path planning controller for multiple unmanned systems (MUS) based on the leader-follower structure. We proposed a combination of global and local planning algorithm and designed the algorithm details for the leader and the follower respectively. The leader was designed to complete path planning task by using a hybrid planning method. Specifically, we chose the jump point search (JPS) algorithm removing some redundant outputs as the global path-finding algorithm, and the deep Q network (DQN) algorithm as the local algorithm. For the follower, the DQN algorithm is adopted to keep appropriate distances with the leader. And the details of the DQN algorithm are designed separately. Finally, we conducted experiments in the Gazebo simulation environment The results show that the framework is executable and robust and it can enable MUS to avoid collisions and achieve the end point