AdvSim: Generating Safety-Critical Scenarios
for Self-Driving Vehicles


In Conference on Computer Vision and Pattern Recognition (CVPR), 2021

Overview of our proposed adversarial scenario generation pipeline. Our goal is to perturb the maneuvers of interactive actors in an existing scenario with adversarial behaviors that cause realistic autonomy system failures. Given an existing scenario and its original sensor data, we perturb the scenario and update accordingly how the SDV would observe the LiDAR sensor data based on the new scene configuration. We then evaluate the autonomy system on the modified scenario, compute an adversarial objective, and update the proposed perturbation using a search algorithm.


Abstract

As self-driving systems become better, simulating scenarios where the autonomy stack may fail becomes more important. Traditionally, those scenarios are generated for a few scenes with respect to the planning module that takes ground-truth actor states as input. This does not scale and cannot identify all possible autonomy failures, such as perception failures due to occlusion. In this paper, we propose AdvSim, an adversarial framework to generate safety-critical scenarios for any LiDAR-based autonomy system. Given an initial traffic scenario, AdvSim modifies the actors' trajectories in a physically plausible manner and updates the LiDAR sensor data to match the perturbed world. Importantly, by simulating directly from sensor data, we obtain adversarial scenarios that are safety-critical for the full autonomy stack. Our experiments show that our approach is general and can identify thousands of semantically meaningful safety-critical scenarios for a wide range of modern self-driving systems. Furthermore, we show that the robustness and safety of these systems can be further improved by training them with scenarios generated by AdvSim.

BibTeX
@article{Wang2021AdvSim,
    author = {Jingkang Wang and Ava Pun and James Tu and Sivabalan Manivasagam and Abbas Sadat and Sergio Casas and Mengye Ren and Raquel Urtasun},
    title = {AdvSim: Generating Safety-Critical Scenarios for Self-Driving Vehicles},
    journal = {Conference on Computer Vision and Pattern Recognition (CVPR)},
    year = {2021},
}
Text citation

Jingkang Wang, Ava Pun, James Tu, Sivabalan Manivasagam, Abbas Sadat, Sergio Casas, Mengye Ren, and Raquel Urtasun. AdvSim: Generating Safety-Critical Scenarios for Self-Driving Vehicles. In Conference on Computer Vision and Pattern Recognition (CVPR), 2021.


Video Citation
@inproceedings{wang2021advsim,
  author    = {Wang, Jingkang and Pun, Ava and Tu, James and Manivasagam, Sivabalan and Sadat, Abbas and Casas, Sergio and Ren, Mengye and Urtasun, Raquel}
  title     = {AdvSim: Generating Safety-Critical Scenarios for Self-Driving Vehicles},
  booktitle = {{CVPR}},
  pages     = {9909--9918},
  year      = {2021}
}