Distributed Trajectory Generation for Multi-Quadrotors Using Receding Horizon Control and Sequential Convex Programming

Abstract

This paper presents a distributed trajectory planning method supporting parallel computation based on receding horizon control (RHC) and sequential convex programming (SCP) for quadrotor swarms in known environments with obstacles. The proposed method, denoted as distributed RHC-SCP (dRHC-SCP), divides the swarm trajectory planning problem into a series of short-horizon planning problems to reduce the computation burden. In each planning horizon, dRHC-SCP solves the swarm trajectory planning problem in an iterative framework via efficient SCP algorithm. In the iterative process of SCP, dRHC-SCP uses the trajectories generated in the last iteration as the nominal trajectories for next iteration to achieve distributed planning and decoupling of the inter-quadrotor collision avoidance constraints. Simulation studies on several scenarios verify the efficiency merit of dRHC-SCP. Comparative results with decoupled SCP (dSCP) demonstrate that dRHC-SCP has higher computational efficiency and better scalability for quadrotor swarm trajectory planning.