An Intelligent System for Generation of Automatic Flight Operation Plans for the Satellite Control Activities at INPE
Luciana Sêda Cardoso, Mauricio Goncalves Vieira Ferreira, Valcir Orlando · SpaceOps 2006 Conference · 2006
Research of automatic solutions for space operations is a real need for all space agencies in order to reduce space mission costs.Nowadays, a significant parcel of satellite operation activities at the National Institute for Space Research (INPE) is still performed manually.Thus, finding automated alternatives for the satellite operation activities at INPE is of capital importance, in order to maintain the currently satisfactory performance of these activities, despite the scarcity of financial resources.This paper proposes the architecture of an Intelligent Planning System for the automatic generation of satellite flight operation plans (PlanIPOV).The proposed system employs temporal planning techniques of artificial intelligence (IA) in the automatic flight operation plans (FOP) generation for a satellite routine operational phase, with the aim of opening the way toward a higher degree of automation for satellite operation activities of INPE.The main reason for the application of the planning system in the routine operational phase of the satellite lifespan is that this phase is composed of very repetitive and well defined tasks which have lower programming costs.In addition, this phase is the longest comprising practically the entire lifespan of the satellite.The PlanIPOV system uses the Planning Domain Definition Language (PDDL2.2) to model the knowledge base of INPE satellite operations.It is based on the automatic generation of problem files, i.e., the initial state of the satellite control environment and the goal to be reached by executing the generated FOP timeline.The system uses information extracted from the following files: tracking knowledge domain; prediction of future satellite passes of the involved ground stations; and configuration parameters of the satellites and the ground stations.This paper also presents a prototype of the proposed planning system.This prototype was implemented using the PDDL2.2language and LPG-TD planner (Local Search for Planning Graphs -Time Initial Literal and Derived Predicates).It was tested for the tracking domain of the satellites currently being controlled by INPE (SCD1, SCD2 and CBERS2).These results are presented along with the limitations that have been observed in the planning technique application.The solutions adopted to overcome these limitations are outlined.The obtained results may be considered satisfactory.