Scientists Explore Optimal Linear Rendezvous Using Reinforcement Learning

In a research paper recently published in Space: Science & Technology, researchers from Harbin Institute of Technology proposed a reinforcement learning-based approach to design the multi-impulse rendezvous trajectories in linear relative motions. This approach enables the rapid generation of rendezvous trajectories through the offline training and the on-board deployment. The mathematical model describing the multi-impulse linear rendezvous problem and the RL algorithms used are provided, and the RL-based approach to rendezvous design is presented. The objective function is the total velocity increment for the fuel-optimal orbital rendezvous problem, while the objective function is the rendezvous time for the time-optimal orbital rendezvous problem.