Probabilistic optimization applied to spacecraft rendezvous and docking

Abstract

This paper presents a probabilistic approach to solve optimal control problems with application to spacecraft proximity operations. The 6 degree-of-freedom rendezvous and docking problem using impulsive control, and avoidance of known obstacles and plume impingement is solved. Our solution is then extended to real-time obstacle avoidance. The space is searched by expanding from the start location by applying only feasible controls and coasts, reducing by nearly 50% the variables perturbed in the search. A randomized A* expansion technique explores the search space. A gradient descent approach smoothes the path and avoids new obstacles in real-time by "stretching" the best pre-computed path in a locally optimal manner.