Aug 2026
PRIMAL3: Pathfinding via Reinforcement and Imitation Multi-Agent Learning - Leveraging LaCAM3
The resulting framework combines learned exploration with structured expert guidance without requiring LaCAM3 at inference and substantially outperforms state-of-the-art learning-based baselines and scales to ultra-large instances with up to city-level 100,000 agents.
Cheng-Yang He, T. Duhan, Gadiel Sznaier Camps et al.
· arXiv.org · 1 citation