Decentralized privacy-preserving onboard mission planning for multi-probe system. (February 2021)

• Record Type: Journal Article
• Authors: Zhao, Yuting; Xu, Rui; Jiang, Huiping; Li, Zhaoyu; Zhu, Shengying; Liang, Zixuan
• Is Part Of: Acta astronautica. Volume 179(2021)
• Page Start: 130
• Page End: 145
• Publication Date: 2021-02
• Subjects: Multi-probe system -- Multi-agent decentralized planning -- Privacy-preserving planning -- Complex constraints; Astronautics -- Periodicals; Outer space -- Exploration -- Periodicals; Astronautics; Periodicals; 629.405
• DOI: 10.1016/j.actaastro.2020.10.041
• Languages: English
• ISSNs: 0094-5765

Abstract

Abstract: Complex deep-space missions containing several tasks are beyond the capacity of a single probe and require multi-probe cooperation. Such missions need onboard planning to avoid communication delays caused by the long distance between the probes and the Earth. However, the traditional centralized planning approaches will burden the manager probe with a high computing load and make the system liable to a single-point failure. A decentralized onboard planning system based on privacy-preserving multi-agent negotiation is proposed to address this problem. The Privacy-preserving Decentralized Multi-Agent Mission Planning Frame (PDMA-MPF) is designed, in which the concepts of private agent and public agent are proposed for the first time. Based on the planning frame, the Decentralized Multi-Agent Iterative Negotiation Task Allocation Method (DMAIN) is developed to allocate tasks and generate local plans for probes. The method uses a deferred-acceptance negotiation pattern, considering the mission profit and load balancing of probes. An information consistency mechanism is designed for the public data synchronization of planners. The local planning of probes is modeled as a constraint satisfaction problem with profit optimization. A temporal constraint handling algorithm and a Resource Rate Timeline (RRTL) based resource constraint handling algorithm are proposed to satisfy constraints in a profit optimized way. The planning system verification shows that the proposed system can generate reasonable plans and guarantee public data consistency between planners. The simulation results illustrate that the proposed planning system gets higher mission profits and a more balanced workload of probes, as compared with the traditional Hierarchical Greedy Algorithm. Highlights: A planning system for a multi-probe system in deep-space missions is proposed. A decentralized privacy-preserving multi-agent planning frame is introduced. Multi-agent negotiation allocates tasks optimally and balances the workload. Complex time and resource constraints of multi-payload probes are handled. The planning system performs well under different task and probe numbers.