Multi-Agent Imitation Learning for Driving Simulation
This work addresses the challenge of realistic multi-agent simulation for autonomous vehicle safety validation, though it is incremental as it builds on existing GAIL methods.
The paper tackled the problem of single-agent imitation learning models failing to generalize to multi-agent driving scenarios by extending Generative Adversarial Imitation Learning (GAIL) with a parameter-sharing approach based on curriculum learning, resulting in policies that interact more stably and capture emergent human driver behavior better than single-agent GAIL.
Simulation is an appealing option for validating the safety of autonomous vehicles. Generative Adversarial Imitation Learning (GAIL) has recently been shown to learn representative human driver models. These human driver models were learned through training in single-agent environments, but they have difficulty in generalizing to multi-agent driving scenarios. We argue these difficulties arise because observations at training and test time are sampled from different distributions. This difference makes such models unsuitable for the simulation of driving scenes, where multiple agents must interact realistically over long time horizons. We extend GAIL to address these shortcomings through a parameter-sharing approach grounded in curriculum learning. Compared with single-agent GAIL policies, policies generated by our PS-GAIL method prove superior at interacting stably in a multi-agent setting and capturing the emergent behavior of human drivers.