Minimizing Coordination in Multi-Agent Path Finding with Dynamic Execution

Abstract

Multi-agent Path Finding (MAPF) is an important problem in large games with many dynamic agents that need to follow space-time trajectories without inter-agent collisions. Modern MAPF solvers plan assuming that agents directly follow the space-time trajectories at known constant speeds without delays or speedups, resulting in rigid plans which need to be replanned if there are changes during execution. Instead we would like agents to be able to follow their computed paths with dynamic velocities while requiring minimal coordination with others to prevent collisions and deadlocks. One way to address this problem is to first produce collision free space-time paths and then compute a coordination controller that prevents collisions and deadlock during dynamic execution. This two step process prevents fully minimizing coordination as the initially planned space-time paths do not reason about coordination and can be arbitrarily bad. We introduce a novel paradigm and show how planning in space-coordination level, rather than space-time, allows us to simultaneously plan paths and a coordination controller. Our method, Space-Level Conflict-Based Search (SL-CBS), builds on the Conflict-Based Search framework and allows us to reason explicitly about coordination, producing paths as well as a coordination controller with bounded suboptimal minimal coordination. We show experimentally that this results in a 20-50% reduction in coordination compared to the closest state of the art solver.

Bidirectional Temporal Plan Graph: Enabling Switchable Passing Orders for More Efficient Multi-Agent Path Finding Plan Execution

“…The Thirty-Eighth AAAI Conference on Artificial Intelligence (AAAI-24) to unnecessary waits. One way to reduce such unnecessary waits is to find MAPF plans that explicitly minimize coordination, such as SL-CBS (Wagner, Veerapaneni, and Likhachev 2022) and DBS (Okumura et al 2023). However, these models usually impose overly strict coordination constraints, making them more challenging to solve than regular MAPF and resulting in longer execution times.…”

From Space-Time to Space-Order: Directly Planning a Temporal Planning Graph by Redefining CBS (Extended Abstract)

“…Once the space-time paths are computed, the TPG post-processing is fixed. Prior work (Wagner, Veerapaneni, and Likhachev 2022) shows that the independently computed space-time paths can result in TPGs that require a large amount of coordination. Instead, if we could directly plan a TPG that agents could follow, we could directly reason about and minimize the amount of coordination that the agents require.…”