Judgelight: Trajectory-Level Post-Optimization for Multi-Agent Path Finding via Closed-Subwalk Collapsing

TL;DR

Judgelight is a post-optimization method that refines multi-agent trajectories by collapsing redundant movements, significantly improving solution quality for MAPF problems, especially when combined with learning-based solvers.

Contribution

It introduces Judgelight, a novel post-optimization layer that formalizes closed-subwalk collapsing as an NP-hard problem and provides an ILP solution to enhance trajectory quality.

Findings

Reduces solution cost by around 20%

Effectively improves trajectories from learning-based MAPF solvers

Maintains all feasibility constraints during optimization

Abstract

Multi-Agent Path Finding (MAPF) is an NP-hard problem with applications in warehouse automation and multi-robot coordination. Learning-based MAPF solvers offer fast and scalable planning but often produce feasible trajectories that contain unnecessary or oscillatory movements. We propose Judgelight, a post-optimization layer that improves trajectory quality after a MAPF solver generates a feasible schedule. Judgelight collapses closed subwalks in agents' trajectories to remove redundant movements while preserving all feasibility constraints. We formalize this process as MAPF-Collapse, prove that it is NP-hard, and present an exact optimization approach by formulating it as integer linear programming (ILP) problem. Experimental results show Judgelight consistently reduces solution cost by around 20%, particularly for learning-based solvers, producing trajectories that are better suited for real-world deployment.