On the application of least-commitment and heuristic search in temporal planning

Graphplan planning graphs are structures widely used in modern planners. The exclusion relations calculated in the planning graph extension provide very useful information, especially in temporal planning where actions have different duration. However, Graphplan backward search has some inefficiencies that impose limitations when dealing with large temporal problems. This paper presents a new search process for temporal planning to avoid these inefficiencies. This search uses the information of a planning graph and shows beneficial in the scalability of the planner. Moreover, our experiments show that a planner with this new search is competitive with other state-of-the-art planners w.r.t. the plan quality.