APPLYING AN AI PLANNER TO SCHEDULE GROUND SATELLITES OPERATIONS

Planning and scheduling are closely related areas that AI community has been long concerned with. The first one deals with finding plans to achieve some goals from an initial state. The second one refers to the allocation of available resources to known activities over time in order to produce schedules that respect temporal relations and resource capacity constrains. They can also optimize a set of objectives, such as minimize tardiness, minimize work in process, maximize resource allocation or minimize cycle time. Planning refers to what should be done and scheduling to how it should be done. Due to these differences, planning and scheduling have been traditionally solved separately using rather different methods. Depending on the problem complexity some domains allow a strict separation between planning and scheduling. But, in other cases, there is an indirect temporal and resource dependency with other states and goals that can not be taken into account if we separate both tasks. In this paper we present a way of integrating them that consists on allowing the planner to reason about resources, although not as explicitly as in the case of other systems. We wanted to explore how to use a standard planner, PRODIGY, to also schedule in a real domain. In this approach, the designer has to be able of determining the elements that can and can not be tackled and the way to express them in the planner language for an efficient internal treatment. In PRODIGY there is not a language that allows an explicit representation of resource or temporal information as in [Chien et al 2000]. However, thanks to three representation facts, the user can use numeric variables and constrains. They are: its capability to represent infinite types (numeric variables); the possibility to define functions that obtain variables values in preconditions of operators; and the use of control rules to prune the search. We have used this approach to plan and schedule nominal operations to perform in three satellites along the year for a Spanish satellites company, HISPASAT. Satellite Operations