Generating Useful Accident-Prone Driving Scenarios via a Learned Traffic Prior
This addresses the need for scalable scenario generation to evaluate and improve autonomous vehicle planners, though it is incremental as it builds on existing traffic modeling and optimization techniques.
The paper tackles the problem of generating realistic and challenging driving scenarios for autonomous vehicle planning by introducing STRIVE, which uses a learned traffic model to create accident-prone scenarios that cause collisions with a given planner, and shows successful generation for two planners and hyperparameter optimization.
Evaluating and improving planning for autonomous vehicles requires scalable generation of long-tail traffic scenarios. To be useful, these scenarios must be realistic and challenging, but not impossible to drive through safely. In this work, we introduce STRIVE, a method to automatically generate challenging scenarios that cause a given planner to produce undesirable behavior, like collisions. To maintain scenario plausibility, the key idea is to leverage a learned model of traffic motion in the form of a graph-based conditional VAE. Scenario generation is formulated as an optimization in the latent space of this traffic model, perturbing an initial real-world scene to produce trajectories that collide with a given planner. A subsequent optimization is used to find a "solution" to the scenario, ensuring it is useful to improve the given planner. Further analysis clusters generated scenarios based on collision type. We attack two planners and show that STRIVE successfully generates realistic, challenging scenarios in both cases. We additionally "close the loop" and use these scenarios to optimize hyperparameters of a rule-based planner.