kei18 / pibt

Priority Inheritance with Backtracking for Iterative Multi-agent Path Finding (IJCAI-19)

PIBT

Let multiple agents move smoothly. See the project page.

IMPORTANT NOTES

• In v1.0-v1.2, the PPS implementation includes bugs. Use the latest (≥v1.3). You can check the PPS results here. The results are not so different from the PIBT paper.
• A new version (pibt2), substantially updated for a journal paper, is out. I strongly recommend using the new one instead of this repo.

Demo

Multi-agent Path Finding

Multi-agent Pickup and Delivery (sushi-mode)

Requirement

The visualization relies on openFrameworks. You need to install openFrameworks beforehand and export OF_ROOT of your environment.

export OF_ROOT={your openFrameworks directory}

The latest implementation relies on boost.

Usage

At first, you must prepare param file. To confirm details, see a sample file. Then you can execute the simulator as following.

• implementation with openFrameworks

make of
make ofrun param=sample-param.txt

• without visualization (for experiment)

make c
make crun param=sample-param.txt

Licence

This software is released under the MIT License, see LICENSE.txt.

Others

• Maps: https://www.movingai.com/benchmarks/grids.html
• Images: https://www.irasutoya.com
• Font: https://fonts.google.com/
• Scenario: https://www.movingai.com/benchmarks/mapf/index.html

Author

Keisuke Okumura is currently a Ph.D. candidate at Tokyo Institute of Technology, working on multiple moving agents.