Symphony: Learning Realistic and Diverse Agents for Autonomous Driving Simulation
This work addresses the need for more accurate simulation agents to accelerate autonomous vehicle development, representing a strong incremental improvement over existing learning from demonstration methods.
The paper tackles the problem of generating realistic and diverse human road user models for autonomous driving simulation by proposing Symphony, which combines conventional policies with a parallel beam search and a hierarchical goal-based approach, resulting in agents that outperform baselines in realism and diversity on proprietary and open datasets.
Simulation is a crucial tool for accelerating the development of autonomous vehicles. Making simulation realistic requires models of the human road users who interact with such cars. Such models can be obtained by applying learning from demonstration (LfD) to trajectories observed by cars already on the road. However, existing LfD methods are typically insufficient, yielding policies that frequently collide or drive off the road. To address this problem, we propose Symphony, which greatly improves realism by combining conventional policies with a parallel beam search. The beam search refines these policies on the fly by pruning branches that are unfavourably evaluated by a discriminator. However, it can also harm diversity, i.e., how well the agents cover the entire distribution of realistic behaviour, as pruning can encourage mode collapse. Symphony addresses this issue with a hierarchical approach, factoring agent behaviour into goal generation and goal conditioning. The use of such goals ensures that agent diversity neither disappears during adversarial training nor is pruned away by the beam search. Experiments on both proprietary and open Waymo datasets confirm that Symphony agents learn more realistic and diverse behaviour than several baselines.