ROS2-based multi-robot warehouse simulation for fault-tolerant coordination research
The project provides a full ROS2 and Gazebo environment to study decentralized, fault‑tolerant coordination of AMR fleets in a warehouse setting. It implements three coordination modes—rule‑based auction, local LLM reasoning, and a hybrid of both—plus a failover mechanism when the central manager crashes. Researchers can inject faults, run long‑horizon experiments, and collect reproducible metrics, making it a valuable testbed for robotics and AI studies. Its modular design and open‑source code let users extend or adapt the platform to new scenarios without starting from scratch.
View on GitHub →Pouya-Mansournia/warehouse-amr-emergent-agents-public