Learning Control Admissibility Models with Graph Neural Networks for Multi-Agent Navigation
This addresses the problem of exponential interaction complexity in multi-agent navigation for robotics or simulation domains, offering a scalable solution that eliminates reward engineering, though it is incremental in its method adaptation.
The paper tackles the challenge of multi-agent navigation in dense environments by shifting from learning optimal actions to predicting sets of admissible actions using control admissibility models (CAMs) with graph neural networks, achieving better performance than state-of-the-art methods when scaled to hundreds of agents.
Deep reinforcement learning in continuous domains focuses on learning control policies that map states to distributions over actions that ideally concentrate on the optimal choices in each step. In multi-agent navigation problems, the optimal actions depend heavily on the agents' density. Their interaction patterns grow exponentially with respect to such density, making it hard for learning-based methods to generalize. We propose to switch the learning objectives from predicting the optimal actions to predicting sets of admissible actions, which we call control admissibility models (CAMs), such that they can be easily composed and used for online inference for an arbitrary number of agents. We design CAMs using graph neural networks and develop training methods that optimize the CAMs in the standard model-free setting, with the additional benefit of eliminating the need for reward engineering typically required to balance collision avoidance and goal-reaching requirements. We evaluate the proposed approach in multi-agent navigation environments. We show that the CAM models can be trained in environments with only a few agents and be easily composed for deployment in dense environments with hundreds of agents, achieving better performance than state-of-the-art methods.