Reproduce “MAPF-HD: Multi-Agent Path Finding in High-Density Environments” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Hiroya Makino, Seigo Ito
Toyota Central R&D Labs researchers developed PHANS (Phased Null-Agent Swapping), an algorithm for Multi-Agent Path Finding in high-density environments (MAPF-HD) that efficiently plans collision-free paths for target agents while relocating obstructing agents. The method successfully solves large-scale problems with 95% agent occupancy within seconds, a significant speed improvement over computationally intensive optimal solvers.
Key insights
- A phased approach, separating target path planning from the detailed management of obstructing agent movement, significantly enhances computational scalability for high-density MAPF.
- Incorporating an implicit cost for obstructing agent relocation within the A* heuristic during target path planning can guide the algorithm towards more efficient global solutions.
- The 'null-agent swapping' mechanism provides a practical and effective way to locally resolve blockages caused by obstructing agents, allowing target agents to advance.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.