Ascent Trajectory Planning for Combined-Power Reusable Launch Vehicles Based on the DDPG Algorithm

This paper proposes an ascent phase trajectory planning method for combined-cycle reusable launch vehicles (CRLV) based on the DDPG algorithm. First, the motion model and flight constraint model for the vehicle's ascent phase were established. Based on these constraints and flight characteristics, a Markov Decision Process (MDP) was designed using the DDPG algorithm to meet flight requirements, addressing the online trajectory planning problem for the CRLV. The vehicle can calculate the optimal control values in real time based on its current motion state. Finally, by comparing the optimization results of the DDPG algorithm and the Particle Swarm Optimization (PSO) algorithm, the superiority and effectiveness of the proposed method were fully demonstrated.