Operations research and artificial intelligence cooperation to solve scheduling problems: the OPAL and OSCAR systems

The two systems described are hybrid systems making OR and AI cooperate to solve scheduling problems. These realizations prove the validity of the global approach which allows one to build efficient and flexible software: operations research builds a relevant search tree and guarantees strong constraints satisfaction, whereas artificial intelligence techniques allow the solving process to make decisions according to domain-dependent rules the user can write. Moreover, real-life applications (job-shop scheduling and mission planning) have been implemented and prove that this approach is operational. A general architecture for such problems has therefore been proposed, which may be extended to all the problems for which theory-based methods already exist.<
>