This code is a combination of Decision Causal Communication (DCC) and the hybrid policy MAPF framework, R$^3$.
The distance heat map generated by Dijkstra algorithm has been used in DCC to determine heuristic guidance in the observation space. When no other agents are in the FOV, this code uses the heat map established in DCC instead of the RRA$^*$ algorithm to select the optimal action.
No. | Folder Name | Description |
---|---|---|
1 | DCC_with_R3 | Code of DCC with R$^3$. |
2 | Generate_Instance | Generate MAPF instances for testing. |
3 | environment.yml | Project environment configuration. |
# python
# Create project environment
conda env create -f environment.yml
@article{ma2021learning,
title={Learning selective communication for multi-agent path finding},
author={Ma, Ziyuan and Luo, Yudong and Pan, Jia},
journal={IEEE Robotics and Automation Letters},
volume={7},
number={2},
pages={1455--1462},
year={2021},
publisher={IEEE}
}