Completable scheduling: An integrated approach to planning and scheduling
Melinda T. Gervasio, GERALD F. DEJONG · 1992
The plannln g problem has traditionally been treated separately from the scheduling problem. However, as more realistic do-mains are tackled, it becomes evident that the problem of de-ciding on an ordered set of tasks to achieve a set of goals cannot be treated independently of the problem of actually allocating resources to the tasks. Doing so would result in losing the ro-bustness and flexibility needed to deal with imperfectly mod-eled domains. Completable scheduling is an approach which integrates the two problems by allowing an a priori planning module to defer particular planning decisions, and conse-quently the associated scheduling decisions, until execution time. This allows a completable scheduling systemto maxi-mize plan flexibility by allowing rantime information to be taken into consideration when making planning and schedul-ing decisions. Furthermore, through the criterion of achievab-ility placed on deferred decisions, a completable scheduling system is able to retain much of the goal--directedness and guarantees of achievement affordedby a priori planning. The eompletable scheduling approach is further enhanced by the use of contingent explanation--based learning, which enables a completable scheduling system to learn general completable plans from example and improve its performance through ex-perience. Initial experimental results show that completable scheduling outperforms classical scheduling as well as pure reactive scheduling in a simple scheduling domain.