Boost the Integration of Planning and Scheduling: a Heuristics Approach

Abstract

Temporal planning embodies aspects of both planning and scheduling. Many temporal planners handle these two subproblems in a loose coupling way. This way simplifies the temporal planning problem but restricts the modeling power. In particular, the simplification fails to handle such temporal planning problems that require concurrency, where actions must execute concurrently to achieve expected effects. For those temporally expressive planning problems, the problem of how to integrate planning with scheduling is emphasized for the sake of both finding a valid plan and further, in an effective way. This paper examines three factors that affect the integrated system's efficiency: information sharing, computation burden balance and interaction frequency. The approach attributes to designing a set of heuristics. By conducting preliminary experiments, the results show good performance of those heuristics compared with the start-of-the-art planner VHPOP.