RWARE (Robotic Warehouse)

Cooperative multi-agent shelf delivery in a simulated warehouse.

Install:

pip install -e 3rd_party/robotic-warehouse/

Paradigm:

Multi-agent (simultaneous)

Stepping:

SIMULTANEOUS

Environment	Description
rware-tiny-2ag/4ag-v2	Tiny warehouse (2 or 4 agents)
rware-small-2ag/4ag-v2	Small warehouse
rware-medium-2ag/4ag-v2	Medium warehouse (also easy/hard reward variants)
rware-large-4ag/8ag-v2	Large warehouse (also hard reward variants)

Citation

@inproceedings{papoudakis2021rware,
  author       = {Georgios Papoudakis and Filippos Christianos and Lukas Sch{\"a}fer and Stefano V. Albrecht},
  title        = {Benchmarking Multi-Agent Deep Reinforcement Learning Algorithms in Cooperative Tasks},
  booktitle    = {Proceedings of the Neural Information Processing Systems Track on Datasets and Benchmarks (NeurIPS)},
  year         = {2021},
}