Spacecraft Swarm Guidance Using a Sequence of Decentralized Convex Optimizations

This paper presents partially decentralized path planning algorithms for swarms of spacecraft composed of hundreds to thousands of agents with each spacecraft having limited computational capabilities. In our prior work, J2-invariant orbits have been found to provide collision free motion for hundreds of orbits. This paper develops algorithms for the swarm reconfiguration which involves transferring from one J2-invariant orbit to another avoiding collisions and minimizing fuel. To perform collision avoidance, it is assumed that the spacecraft can communicate their trajectories with each other. The algorithm uses sequential convex programming to solve a series of approximate path planning problems until the solution converges. Two decentralized methods are developed: a serial method where the spacecraft take turn updating their trajectories and a parallel method where all of the spacecraft update their trajectories simultaneously.