Interaction-Aware Multi-Agent Reinforcement Learning for Mobile Agents with Individual Goals
This work addresses the problem of decentralized learning in non-stationary environments for mobile agents with individual goals, representing an incremental advancement in multi-agent reinforcement learning.
The paper tackles the challenge of training non-stationary policies in multi-agent reinforcement learning for mobile robot navigation by identifying mutual adaptation to sub-optimal behaviors as a key difficulty and proposing a two-stage curriculum-based strategy. The approach was evaluated on autonomous driving and robot navigation domains, showing improved performance in decentralized learning settings.
In a multi-agent setting, the optimal policy of a single agent is largely dependent on the behavior of other agents. We investigate the problem of multi-agent reinforcement learning, focusing on decentralized learning in non-stationary domains for mobile robot navigation. We identify a cause for the difficulty in training non-stationary policies: mutual adaptation to sub-optimal behaviors, and we use this to motivate a curriculum-based strategy for learning interactive policies. The curriculum has two stages. First, the agent leverages policy gradient algorithms to learn a policy that is capable of achieving multiple goals. Second, the agent learns a modifier policy to learn how to interact with other agents in a multi-agent setting. We evaluated our approach on both an autonomous driving lane-change domain and a robot navigation domain.