Agent Modelling under Partial Observability for Deep Reinforcement Learning
This addresses the challenge of partial observability in multi-agent systems for AI researchers, though it is an incremental improvement over existing methods.
The paper tackles the problem of agent modeling in multi-agent reinforcement learning without requiring access to other agents' observations and actions during execution, by learning representations from the controlled agent's local observations and using them to augment its policy, achieving higher returns than baselines in various environments.
Modelling the behaviours of other agents is essential for understanding how agents interact and making effective decisions. Existing methods for agent modelling commonly assume knowledge of the local observations and chosen actions of the modelled agents during execution. To eliminate this assumption, we extract representations from the local information of the controlled agent using encoder-decoder architectures. Using the observations and actions of the modelled agents during training, our models learn to extract representations about the modelled agents conditioned only on the local observations of the controlled agent. The representations are used to augment the controlled agent's decision policy which is trained via deep reinforcement learning; thus, during execution, the policy does not require access to other agents' information. We provide a comprehensive evaluation and ablations studies in cooperative, competitive and mixed multi-agent environments, showing that our method achieves higher returns than baseline methods which do not use the learned representations.