abmarl 0.1.1

Agent Based Simulation and MultiAgent Reinforcement Learning

Abmarl

Abmarl is a package for developing agent-based simulations and training them with multiagent reinforcement learning. We provide an intuitive command line interface for training, visualizing, and analyzing agent behavior. We define an Agent Based Simulation Interface and Simulation Managers, which control which agents interact with the simulation at each step. We support integration with several popular simulation interfaces, including gym.Env and MultiAgentEnv.

Abmarl is a layer in the Reinforcement Learning stack that sits on top of RLlib. We leverage RLlib’s framework for training agents and extend it to more easily support custom simulations, algorithms, and policies. We enable researchers to rapidly prototype RL experiments and simulation design and lower the barrier for pre-existing projects to prototype RL as a potential solution.

Getting started

To use Abmarl, install via pip: pip install abmarl

To develop Abmarl, clone the repository and install via pip's development mode:

git clone [e-mail]:LLNL/Abmarl.git
cd abmarl
pip install -r requirements.txt
pip install -e . --no-deps

Train agents in a multicorridor simulation:

abmarl train examples/multi_corridor_example.py

Documentation

You can find the latest Abmarl documentation, on our ReadTheDocs page.

Contact

• Edward Rusu, [e-mail]
• Ruben Glatt, [e-mail]

Release

LLNL-CODE-815883