Multi-Agent Robot Coordination
Research
Cooperative cleaning robots trained with multi-agent RL. The code behind my Springer LNNS paper.
PythonPyTorchStable-Baselines3PPOPygame
A simulation of cooperative cleaning robots sharing a room, used to compare centralised and decentralised multi-agent reinforcement learning.
- Centralised PPO and joint DQN via Stable-Baselines3; IDQN, MAPPO and QMIX implemented from scratch in PyTorch.
- Tracks room coverage and agent steps, with visualisations of policy behaviour at different training checkpoints.
- Published as Multi-agent Reinforcement Learning for Autonomous Mobile Robot Coordination (IAT 2026, Springer LNNS 2278). See publications.