부산대학교 도서관

Space observation has become one of the most common fields for showcasing planning and scheduling technology. This paper describes the solutions adopted in the solver of the Sensor Planning System, part of the Space Situational Awareness (SSA) Preparatory Programme of the European Space Agency (ESA). The goal of this solver is to find a feasible conflict-free assignment of a set of objects to be tracked by a set of sensors, given several constraints. We use an off-the-shelf temporal planner to solve this problem, Optic. A key aspect of this task is that problems are usually huge. Thus, current temporal planners can not solve those problems directly. We explain first how to model the problem as a temporal planning problem, and then we present an approach for partitioning the problem in order to be able to solve it that allows a multi-core processing. The experimental results, both with simulated and real data, confirm the approach is useful for solving target problems of realistic size.