Realization of multi-agent planning system for autonomous spacecraft

To achieve fully autonomy of deep space spacecraft, onboard planning is a crucial technology. Though, lots of successful applications about it in spacecraft operations have recently been used in real or experimental flight, the inherent distribution and concurrency of the spacecraft subsystem was not taken into consideration. To improve the efficiency of planning system, a multi-agent planning system (MAPS) of autonomous spacecraft is proposed in this paper. MAPS subdivides the search space into some small ones and every PA manages and operates in its own space. With the help of planning manager agent (PMA), all of the planning agents communicate and cooperate with each other to produce partial plan which satisfied all the constraints of the system. To be capable of describing simultaneous activity, continue time, resource and temporal constraints in MAPS, a new planning formal model is given firstly. As the application of this architecture, an MAPS prototype system for deep spacecraft, which is realized with Java language, is designed.