Publication

R. Barak, N. Zhou, R. Stern, E. Boyarski and P. Surynek. Modeling and Solving the Multi-Agent Pathfinding Problem in Picat. In Proceedings of the International Conference on Tools with Artificial Intelligence (ICTAI), pages 959-966, 2017.

Abstract: The multi-agent pathfinding (MAPF) problem has attracted considerable attention because of its relation to practical applications. In this paper, we present a constraint-based declarative model for MAPF, together with its implementation in Picat, a logic-based programming language. We show experimentally that our Picat-based implementation is highly competitive and sometimes outperforms previous approaches. Importantly, the proposed Picat implementation is very versatile. We demonstrate this by showing how it can be easily adapted to optimize different MAPF objectives, such as minimizing makespan or minimizing the sum of costs, and for a range of MAPF variants. Moreover, a Picat-based model can be automatically compiled to several general-purpose solvers such as SAT solvers and Mixed Integer Programming solvers (MIP). This is particularly important for MAPF because some MAPF variants are solved more efficiently when compiled to SAT while other variants are solved more efficiently when compiled to MIP. We analyze these differences and the impact of different declarative models and encodings on empirical performance.

Download the paper in pdf.