mlsanityData to 7 Oct 2026

How Do We Fail? Stress Testing Perception in Autonomous Vehicles

arXiv 2203.14155 ↗First posted 26 Mar 2022cs.RO cs.AI cs.LGcited by 17

On the map: End-to-end autonomous driving → Adversarial scenarios and safety testing

Abstract (arXiv)

Autonomous vehicles (AVs) rely on environment perception and behavior prediction to reason about agents in their surroundings. These perception systems must be robust to adverse weather such as rain, fog, and snow. However, validation of these systems is challenging due to their complexity and dependence on observation histories. This paper presents a method for characterizing failures of LiDAR-based perception systems for AVs in adverse weather conditions. We develop a methodology based in reinforcement learning to find likely failures in object tracking and trajectory prediction due to sequences of disturbances. We apply disturbances using a physics-based data augmentation technique for simulating LiDAR point clouds in adverse weather conditions. Experiments performed across a wide range of driving scenarios from a real-world driving dataset show that our proposed approach finds high likelihood failures with smaller input disturbances compared to baselines while remaining computationally tractable. Identified failures can inform future development of robust perception systems for AVs.

Key facts. Cited by 17 papers in the map (data through 7 Oct 2026). Placed in Adversarial scenarios and safety testing, one of 9 directions in End-to-end autonomous driving. Builds on 24 of its 42 references that are in the map. First posted on arXiv 26 Mar 2022.

Builds on

24 of 42 references.

References · 24

  1. 2019nuScenes: A multimodal dataset for autonomous drivingcited by 4,485
  2. 2018PointPillars: Fast Encoders for Object Detection from Point Cloudscited by 1,707
  3. 2018Threat of Adversarial Attacks on Deep Learning in Computer Vision: A Surveycited by 459
  4. 2017Adversarial Examples: Attacks and Defenses for Deep Learningcited by 350
  5. 2017Adversarial Examples for Semantic Segmentation and Object Detectioncited by 275
  6. 2019CoverNet: Multimodal Behavior Prediction using Trajectory Setscited by 259
  7. 2018Physical Adversarial Examples for Object Detectorscited by 163
  8. 2018Generating 3D Adversarial Point Cloudscited by 144
  9. 2020PnPNet: End-to-End Perception and Prediction with Tracking in the Loopcited by 143
  10. 2018Transferable Adversarial Attacks for Image and Video Object Detectioncited by 109
  11. 2020Lidar for Autonomous Driving: The principles, challenges, and trends for automotive lidar and perception systemscited by 100
  12. 2019Extending Adversarial Attacks and Defenses to Deep 3D Point Cloud Classifierscited by 92
12 more
  1. 2020A Survey of Algorithms for Black-Box Safety Validation of Cyber-Physical Systemscited by 71
  2. 2019Adaptive Stress Testing for Autonomous Vehiclescited by 70
  3. 2019Robustness of 3D Deep Learning in an Adversarial Settingcited by 55
  4. 2021Lidar Light Scattering Augmentation (LISA): Physics-based Simulation of Adverse Weather Conditions for 3D Object Detectioncited by 42
  5. 2020Out-of-Distribution Detection for Automotive Perceptioncited by 37
  6. 2019Deep Learning-based Vehicle Behaviour Prediction For Autonomous Driving Applications: A Reviewcited by 31
  7. 2021Robustness of Object Detectors in Degrading Weather Conditionscited by 28
  8. 2021Fooling LiDAR Perception via Adversarial Trajectory Perturbationcited by 25
  9. 2020Validation of Image-Based Neural Network Controllers through Adaptive Stress Testingcited by 19
  10. 2019Dynamic Environment Prediction in Urban Scenes using Recurrent Representation Learningcited by 18
  11. 2020Double-Prong ConvLSTM for Spatiotemporal Occupancy Prediction in Dynamic Environmentscited by 13
  12. 2020Attention Augmented ConvLSTM for Environment Predictioncited by 12

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2022: 3; 2023: 3; 2024: 9; 2025: 1; 2026: 1.

17 papers.

Sort:

By area:

  1. 2023Survey on LiDAR Perception in Adverse Weather Conditionscited by 23
  2. 2023Scene Informer: Anchor-based Occlusion Inference and Trajectory Prediction in Partially Observable Environmentscited by 14
  3. 2024The Role of Predictive Uncertainty and Diversity in Embodied AI and Robot Learningcited by 11
  4. 2024Failures Are Fated, But Can Be Faded: Characterizing and Mitigating Unwanted Behaviors in Large-Scale Vision and Language Modelscited by 11
  5. 2022Testing Rare Downstream Safety Violations via Upstream Adaptive Sampling of Perception Error Modelscited by 7
  6. 2024Self-supervised Multi-future Occupancy Forecasting for Autonomous Drivingcited by 7
  7. 2022Monitoring of Perception Systems: Deterministic, Probabilistic, and Learning-based Fault Detection and Identificationcited by 6
  8. 2024RoboMD: Uncovering Robot Vulnerabilities through Semantic Potential Fieldscited by 5
  9. 2022LOPR: Latent Occupancy PRediction using Generative Modelscited by 4
  10. 2024Hard Cases Detection in Motion Prediction by Vision-Language Foundation Modelscited by 4
  11. 2024Fairness in Autonomous Driving: Towards Understanding Confounding Factors in Object Detection under Challenging Weathercited by 2
  12. 2024Exploring Critical Testing Scenarios for Decision-Making Policies: An LLM Approachcited by 2
5 more
  1. 2024A novel framework for adaptive stress testing of autonomous vehicles in multi-lane roadscited by 1
  2. 2023ReMAV: Reward Modeling of Autonomous Vehicles for Finding Likely Failure Eventscited by 0
  3. 2024LLM-Assisted Red Teaming of Diffusion Models through "Failures Are Fated, But Can Be Faded"cited by 0
  4. 2025VLC Fusion: Vision-Language Conditioned Sensor Fusion for Robust Object Detectioncited by 0
  5. 2026Scenario Generation for Testing of Autonomous Driving Systems Using Real-World Failure Recordscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Real-World Perturbation Testing of Autonomous Driving Systems
  2. 2025Edge AI-Powered Real-Time Decision-Making for Autonomous Vehicles in Adverse Weather Conditions
  3. 2025Object detection in adverse weather conditions for autonomous vehicles using Instruct Pix2Pix
  4. 2023Adaptive Failure Search Using Critical States from Domain Experts
  5. 2026ObsDriveBench: Benchmarking Multimodal Understanding under Adverse Weather with Observability Awareness
  6. 2026RESBev: Making BEV Perception More Robust
  7. 2023DeepQTest: Testing Autonomous Driving Systems with Reinforcement Learning and Real-world Weather Data
  8. 2022Calibrating Perception Uncertainty for Autonomous Driving