Hint

Example code: https://github.com/cselab/korali/tree/master/examples/reinforcement.learning/vmas/

Multi-Agent Reinforcement Learning with VMAS: A Vectorized Multi-Agent Simulator for Collective Robot Learning

The following examples are based on the VMAS python library. To install please run

pip install vmas

For a description of the MARL tasks and further information we refer to the VMAS github.

Example

For the vmas environments (navigation, balance, sampling) we follow the specifications of TorchRL (3 agents, episode length 100).

The tasks can be selected with the --env flag, e.g. to run the balance task

python run-vracer.py --env balance