An adaptive surrogate model-based fast planning for swarm safe migration along halo orbit

Taken the optimal assignment, trajectory optimization and collision avoidance into consideration, this research focuses on the fast-planning method for swarm safe-migration along Halo orbit. The swarm safe-migration planning problem is an inherently high-dimensional nonlinear problem. Repeatedly solving the bottom layer's collision-free trajectory optimization problem causes heavy computation burden in the traditional two-layer optimization framework. To reduce the computation cost and keep the optimization performance, an adaptive surrogate model-based swarm safe-migration planning method is developed in this paper. The surrogates fitted to the collision-avoidance constraints are investigated and the input variables of the surrogates are analyzed and selected for improved efficiency. In particular, a classifier-based surrogate management strategy is designed for the swarm safe-migration problems to improve the optimization capability of the proposed algorithm. Finally, the proposed method is applied to two real-world swarm migration problems, i.e., a reconfiguration problem with geometric constraint and an assignment optimization problem. Numerical results indicate that, compared with the optimal result from the widely-used Genetic Algorithm method, our method could obtain a similar solution within only about 4% of the computational time.