Train Route Planning as a Multi-agent Path Finding Problem.

The train routing and timetabling problem consists of setting routes and schedules of a set of vehicles given their initial timetables and a railway network. The number of vehicles, the complexity and limited capacity of the railway network, and the time constraints make this problem difficult to solve. In this paper, we model this problem as a Multi-Agent Pathfinding problem, and propose a Conflict-Based Search approach to solve it. In our approach, we consider the complex properties found in this scenario such as continuous time, agents that function as convoys of arbitrary length, arbitrary action duration, and railway networks to find a solution. We analyze and discuss our approach explaining the main difficulties and evaluate it on several scenarios.