Sampling-Based Trajectory Generation for Autonomous Spacecraft Rendezvous and Docking

This paper presents a methodology for computationally ecient, direct trajectory generation for autonomous spacecraft rendezvous using sampling with a distance or time cost function to be minimized. The approach utilizes a randomized A* algorithm called Sampling-Based Model Predictive Optimization (SBMPO) that exclusively samples the input space and integrates the dynamic model of the system. A primary contribution of this paper is the development of appropriate optimistic A* heuristics that take into account a goal position and velocity and are based on the minimum distance and minimum time control problems for the vehicle. These heuristics enable fast computation of trajectories that end in zero relative velocity. Additionally, this paper introduces an alternative approach to collision avoidance for use in graph-based planning algorithms. By referencing the full state of the vehicle, a method of imminent collision detection is applied within the graph search process of the trajectory planner. Using a six degree-of-freedom relative motion spacecraft dynamic model, simulation results are illustrated for generating rendezvous-feasible trajectories in cluttered environments.