GRACE: A Unified 2D Multi-Robot Path Planning Simulator & Benchmark for Grid, Roadmap, and Continuous Environments
Chuanlong Zang, Anna Mannucci, Isabelle Barz, Philipp Schillinger, Florian Lier, Wolfgang Hoenig
AI summary
Problem
Existing multi-robot planning benchmarks are fragmented: grid-based tools scale well but lack realism, while continuous simulators offer fidelity but hinder fair cross-planner comparison. This makes it difficult to answer when simpler abstractions suffice or how different representations affect planning outcomes.
Approach
The authors introduce GRACE, a unified 2D simulator and benchmark that applies explicit, reproducible abstraction operators to map identical tasks across grid, roadmap, and continuous representations, paired with a standardized evaluation protocol and deterministic simulation core.
Key results
- Unified platform with reproducible abstraction operators across grid, roadmap, and continuous representations
- Empirical benchmark comparing planner families on shared instance sets
- Quantified representation-fidelity trade-offs showing continuous models favor fidelity while discrete models scale farther
- Demonstrated scalability up to ~2k agents with deterministic execution
Why it matters
Provides researchers and practitioners with a standardized tool to fairly evaluate and compare multi-robot planning algorithms across abstraction levels, accelerating reliable deployment.
Abstract
Advancing Multi-Agent Pathfinding (MAPF) and Multi-Robot Motion Planning (MRMP) requires platforms that enable transparent, reproducible comparisons across modeling choices. Existing tools either scale under simplifying assump- tions (grids, homogeneous agents) or offer higher fidelity with less comparable instrumentation. We present GRACE, a unified 2D simulator+benchmark that instantiates the same task at multiple abstraction levels (grid, roadmap, continuous) via explicit, reproducible operators and a common evaluation protocol. Our empirical results on public maps and repre- sentative planners enable commensurate comparisons on a shared instance set. Furthermore, we quantify the expected representation–fidelity trade-offs (MRMP solves instances at higher fidelity but lower speed, while grid/roadmap planners scale farther). By consolidating representation, execution, and evaluation, GRACE thereby aims to make cross-representation studies more comparable and provides a means to advance multi-robot planning research and its translation to practice.