Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Engineering LaCAM$^\ast$: Towards Real-Time, Large-Scale, and Near-Optimal Multi-Agent Pathfinding

About

This paper addresses the challenges of real-time, large-scale, and near-optimal multi-agent pathfinding (MAPF) through enhancements to the recently proposed LaCAM* algorithm. LaCAM* is a scalable search-based algorithm that guarantees the eventual finding of optimal solutions for cumulative transition costs. While it has demonstrated remarkable planning success rates, surpassing various state-of-the-art MAPF methods, its initial solution quality is far from optimal, and its convergence speed to the optimum is slow. To overcome these limitations, this paper introduces several improvement techniques, partly drawing inspiration from other MAPF methods. We provide empirical evidence that the fusion of these techniques significantly improves the solution quality of LaCAM*, thus further pushing the boundaries of MAPF algorithms.

Keisuke Okumura• 2023

Related benchmarks

TaskDatasetResultRank
Multi-Agent Path FindingMedium Room 23x23 world size, 33.5% static obstacle rate
Success Rate100
20
Multi-Agent Path FindingSmall Random 10x10 world size, 17.5% static obstacle rate
Success Rate93
20
Multi-Agent Path FindingMedium Maze 25x25 world size, 32.8% static obstacle rate
Success Rate93
20
Multi-Agent Path FindingMedium Warehouse 25x25 world size, 34.6% static obstacle rate
Success Rate100
20
Multi-Agent Path FindingLarge Maze 33x33 world size, 33.0% static obstacle rate
Success Rate44
20
Showing 5 of 5 rows

Other info

Follow for update