Saturated and Consistent Neighborhood for Selecting and Scheduling Photographs of Agile Earth Observing Satellite

The mission of an Agile Earth Observing Satellite (AEOS) is to acquire photographs on the Earth surface, in response to observation requests. The management problem of an AEOS is to select and schedule a subset of weighted images among a set of candidate ones which must satisfy imperative constraints and at the same time maximize a given profit function. This new generation of satellites, like those studied in the French PLEIADES project, are called agiles because of their mobility on three axes (roll, pitch and yaw) while the unique on-board camera remains fixed. This mobility gives rise to a potentially better efficiency of the system. In fact, since the azimuth and the starting time of an image acquisition are free, there is potentially infinite number of possibilities to acquire a given area on the Earth surface. However, managing an AEOS is significantly more difficult due to a very sized search space. The AEOS management problem is NP-hard [5]. To solve it, four methods are investigated in [5]: a fast greedy algorithm, a dynamic programming algorithm, a constraint programming method, and a local search algorithm based on insertion and removal of images in a schedule. The selection and scheduling problem for a non-agile satellite, like SPOT5, is stated in [1] and some methods are dedicated to solve it. The best results are obtained in [8] by using a tabu search, and very good upper bounds are calculated by an original partition method [9]. A scheduling problem involving a satellite equipped with a radar instrument, very agile on the pitch axis but slow on the roll axis, is stated and solved in [3] by a partial enumeration algorithm. In this paper, we propose a tabu algorithm, hybridized with a systematic search in terms of partial enumerations, to solve the problem of selecting and scheduling the photographs of an AEOS. Our tabu search is based on a consistent and saturated neighborhood. This leads to treat only consistent configurations by checking constraint consistency. The paper is organized as follows. Starting by a description of the AEOS management problem, the components of our resolution approach are detailed in Sections 3 to 5. The experimental results are shown in Section 6. Finally, we conclude and give some future works.