Multi-Agent examples

[brean]

Create an example with multiple Agents

• see https://youtu.be/S-VAL7Epn3o?t=588

[nisarganc]

Hello Andreas,

Thank you for the great library! I was wondering if the library already supports multi-agent planning?

[brean]

Hello @nisarganc

Thank you for using it!

it does not directly support it, but you can use it in a multi-agent scenario.

For example each agent could be just an obstacle and you re-plan whenever your robots move. However this could easily lead to problems as re-planning a big environment might take some time, especially with A*, where you need to create the whole path again when an obstacle changes.

Some path caching, for example based on Dijkstra or the implementation of a potential field that can be stored independent of individual agent positions might be a good extension to python-pathfinding for that.

In an environment that is completely explored and where you can pre-plan your agents I recommend the RVO2 library for some applications, see Bermpohl, 2023, Experimental Evaluation of AGV Dispatching Methods in an Agent-Based Simulation Environment and a Digital Twin. For complex multi-agent planning with local obstacle avoidance I recommend a full robotics framework like ROS 2 with nav2 where Theta* is used for the global path planning in combination with DWA as local planner to avoid obstacles like other robots.

If you need it in research feel free to contact me outside of github via mail so we can discuss your exact use case.