Heuristic Search Via Graphical Structure in Temporal Interval-Based Planning for Deep Space Exploration
Hao Jin,Rui Xu,Pingyuan Cui,Shengying Zhu,Huiping Jiang,Feng Zhou
DOI: https://doi.org/10.1016/j.actaastro.2019.10.002
IF: 2.954
2019-01-01
Acta Astronautica
Abstract:Operations of conventional spacecraft used to be planned on ground and are uploaded as telecommands and executed on board at due time. However, because of difficulties in communicating with distant spacecraft, direct human control for the spacecraft is infeasible. Therefore, great hopes are placed on automated planning techniques to enhance the security of the spacecraft. By deciding a coon, as to support opportunistic science, the planner is mplex set of activities or states, an onboard planner is able to effectively arrange the daily tasks on a spacecraft. In additialso required to respond in the shortest possible time. Typically, to better characterize the spacecraft, the timeline-based knowledge representation benefits from its powerful ability to describe time and temporal behaviors, which is essential to effectively address real world problems. Compliant with the representation method, an elegant approach is devised for search guidance and solving problems efficiently in space-like contexts. Specifically, the key technique we build on is the heuristic estimate strategy based on a graphical structure defined in the model. Furthermore, a search algorithm joint with the heuristic function is proposed to avoid redundant work. By evaluating the branching nodes, this approach is able to prune irrelevant search space and make improvements in onboard planning efficiency. Our experiments exhibit an excellent performance on tested instances compared to Europa2.