An Ant System algorithm for automated trajectory planning

The paper presents an Ant System based algorithm to optimally plan multi-gravity assist trajectories. The algorithm is designed to solve planning problems in which there is a strong dependency of one decision on all the previously-made decisions. In the case of multi-gravity assist trajectory planning, the number of possible paths grows exponentially with the number of planetary encounters. The proposed algorithm avoids scanning all the possible paths and provides good results at a low computational cost. The algorithm builds the solution incrementally according to Ant System paradigms. Unlike standard ACO, at every planetary encounter, each ant makes a decision based on the information stored in a tabu and feasible list. The approach demonstrated to be competitive, on a number of instances of a real trajectory design problem, against known GA and PSO algorithms.