mlsanityData to 7 Oct 2026

CAT: Closed-loop Adversarial Training for Safe End-to-End Driving

arXiv 2310.12432 ↗First posted 19 Oct 2023cs.LGcited by 52

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

Abstract (arXiv)

Driving safety is a top priority for autonomous vehicles. Orthogonal to prior work handling accident-prone traffic events by algorithm designs at the policy level, we investigate a Closed-loop Adversarial Training (CAT) framework for safe end-to-end driving in this paper through the lens of environment augmentation. CAT aims to continuously improve the safety of driving agents by training the agent on safety-critical scenarios that are dynamically generated over time. A novel resampling technique is developed to turn log-replay real-world driving scenarios into safety-critical ones via probabilistic factorization, where the adversarial traffic generation is modeled as the multiplication of standard motion prediction sub-problems. Consequently, CAT can launch more efficient physical attacks compared to existing safety-critical scenario generation methods and yields a significantly less computational cost in the iterative learning pipeline. We incorporate CAT into the MetaDrive simulator and validate our approach on hundreds of driving scenarios imported from real-world driving datasets. Experimental results demonstrate that CAT can effectively generate adversarial scenarios countering the agent being trained. After training, the agent can achieve superior driving safety in both log-replay and safety-critical traffic scenarios on the held-out test set. Code and data are available at https://metadriverse.github.io/cat.

Key facts. Cited by 52 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 32 of its 43 references that are in the map. First posted on arXiv 19 Oct 2023.

Builds on

32 of 43 references.

Closest ancestors · 2

Same direction, newest first

  1. 2022SafeBench: A Benchmarking Platform for Safety Evaluation of Autonomous Vehicles20 Jun
  2. 2021Generating Useful Accident-Prone Driving Scenarios via a Learned Traffic Prior9 Dec

Other references · 30

  1. 2016Towards Evaluating the Robustness of Neural Networkscited by 2,976
  2. 2017CARLA: An Open Urban Driving Simulatorcited by 2,431
  3. 2018Addressing Function Approximation Error in Actor-Critic Methodscited by 2,302
  4. 2017Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environmentscited by 1,519
  5. 2020Deep Reinforcement Learning for Autonomous Driving: A Surveycited by 698
  6. 2021Large Scale Interactive Motion Forecasting for Autonomous Driving : The Waymo Open Motion Datasetcited by 532
  7. 2021NuPlan: A closed-loop ML-based planning benchmark for autonomous vehiclescited by 485
  8. 2021DenseTNT: End-to-end Trajectory Prediction from Dense Goal Setscited by 292
  9. 2021MetaDrive: Composing Diverse Driving Scenarios for Generalizable Reinforcement Learningcited by 281
  10. 2022Motion Transformer with Global Intention Localization and Local Movement Refinementcited by 267
  11. 2021MultiPath++: Efficient Information Fusion and Trajectory Aggregation for Behavior Predictioncited by 246
  12. 2021AdvSim: Generating Safety-Critical Scenarios for Self-Driving Vehiclescited by 132
18 more
  1. 2021HOME: Heatmap Output for future Motion Estimationcited by 122
  2. 2020Enhanced POET: Open-Ended Reinforcement Learning through Unbounded Invention of Learning Challenges and their Solutionscited by 103
  3. 2022A Survey on Safety-Critical Driving Scenario Generation -- A Methodological Perspectivecited by 100
  4. 2022KING: Generating Safety-Critical Driving Scenarios for Robust Imitation via Kinematics Gradientscited by 95
  5. 2022On Adversarial Robustness of Trajectory Prediction for Autonomous Vehiclescited by 89
  6. 2022M2I: From Factored Marginal Trajectory Prediction to Interactive Predictioncited by 81
  7. 2020Learning to Collide: An Adaptive Safety-Critical Scenarios Generating Methodcited by 74
  8. 2019PhysGAN: Generating Physical-World-Resilient Adversarial Examples for Autonomous Drivingcited by 72
  9. 2021A Survey of Deep RL and IL for Autonomous Driving Policy Learningcited by 59
  10. 2019Safe Reinforcement Learning on Autonomous Vehiclescited by 48
  11. 2019Failure-Scenario Maker for Rule-Based Agent using Multi-agent Adversarial Reinforcement Learning and its Application to Autonomous Drivingcited by 36
  12. 2020Safe Reinforcement Learning for Autonomous Vehicles through Parallel Constrained Policy Optimizationcited by 34
  13. 2021Safe Driving via Expert Guided Policy Optimizationcited by 29
  14. 2019Improved Robustness and Safety for Autonomous Vehicle Control with Adversarial Reinforcement Learningcited by 23
  15. 2021Learn-to-Race: A Multimodal Control Environment for Autonomous Racingcited by 23
  16. 2021A Survey on Scenario-Based Testing for Automated Driving Systems in High-Fidelity Simulationcited by 21
  17. 2021CausalAF: Causal Autoregressive Flow for Safety-Critical Driving Scenario Generationcited by 20
  18. 2019Simple Physical Adversarial Examples against End-to-End Autonomous Driving Modelscited by 17

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 15; 2025: 16; 2026: 21.

52 papers.

Sort:

By area:

  1. 2024ChatScene: Knowledge-Enabled Safety-Critical Scenario Generation for Autonomous Vehiclescited by 92
  2. 2024HUGSIM: A Real-Time, Photo-Realistic and Closed-Loop Simulator for Autonomous Drivingcited by 84
  3. 2024SimGen: Simulator-conditioned Driving Scene Generationcited by 34
  4. 2024DrivingSphere: Building a High-fidelity 4D World for Closed-loop Simulationcited by 34
  5. 2024CtRL-Sim: Reactive and Controllable Driving Agents with Offline Reinforcement Learningcited by 26
  6. 2025Embodied Scene Understanding for Vision Language Models via MetaVQAcited by 16
  7. 2025LLM-attacker: Enhancing Closed-loop Adversarial Scenario Generation for Autonomous Driving with Large Language Modelscited by 16
  8. 2025Challenger: Affordable Adversarial Driving Video Generationcited by 16
  9. 2024Traffic Scene Generation from Natural Language Description for Autonomous Vehicles with Large Language Modelcited by 14
  10. 2024FREA: Feasibility-Guided Generation of Safety-Critical Scenarios with Reasonable Adversarialitycited by 13
  11. 2025Decoupled Diffusion Sparks Adaptive Scene Generationcited by 12
  12. 2024CaDRE: Controllable and Diverse Generation of Safety-Critical Driving Scenarios using Real-World Trajectoriescited by 11
38 more
  1. 2025CurricuVLM: Towards Safe Autonomous Driving via Personalized Safety-Critical Curriculum Learning with Vision-Language Modelscited by 11
  2. 2024SEAL: Towards Safe Autonomous Driving via Skill-Enabled Adversary Learning for Closed-Loop Scenario Generationcited by 10
  3. 2024ReGentS: Real-World Safety-Critical Driving Scenario Generation Made Stablecited by 8
  4. 2024UrbanCAD: Towards Highly Controllable and Photorealistic 3D Vehicles for Urban Scene Simulationcited by 7
  5. 2025Seeking to Collide: Online Safety-Critical Scenario Generation for Autonomous Driving with Retrieval Augmented Large Language Modelscited by 7
  6. 2026VILTA: A VLM-in-the-Loop Adversary for Enhancing Driving Policy Robustnesscited by 6
  7. 2024Blending Data-Driven Priors in Dynamic Gamescited by 5
  8. 2025RIFT: Group-Relative RL Fine-Tuning for Realistic and Controllable Traffic Simulationcited by 4
  9. 2024Adversarial and Reactive Traffic Entities for Behavior-Realistic Driving Simulation: A Reviewcited by 3
  10. 2025Safety-Critical Traffic Simulation with Adversarial Transfer of Driving Intentionscited by 3
  11. 2026ADV-0: Closed-Loop Min-Max Adversarial Training for Long-Tail Robustness in Autonomous Drivingcited by 3
  12. 2025AccidentSim: Generating Vehicle Collision Videos with Physically Realistic Collision Trajectories from Real-World Accident Reportscited by 2
  13. 2025SceneStreamer: Continuous Scenario Generation as Next Token Group Predictioncited by 2
  14. 2025Optimization-Guided Diffusion for Interactive Scene Generationcited by 2
  15. 2024VCAT: Vulnerability-aware and Curiosity-driven Adversarial Training for Enhancing Autonomous Vehicle Robustnesscited by 1
  16. 2025Towards Benchmarking and Assessing the Safety and Robustness of Autonomous Driving on Safety-critical Scenarioscited by 1
  17. 2025RADE: Learning Risk-Adjustable Driving Environment via Multi-Agent Conditional Diffusioncited by 1
  18. 2025Robust Driving Control for Autonomous Vehicles: An Intelligent General-sum Constrained Adversarial Reinforcement Learning Approachcited by 1
  19. 2025Controllable risk scenario generation from human crash data for autonomous vehicle testingcited by 1
  20. 2025Driving in Corner Case: A Real-World Adversarial Closed-Loop Evaluation Platform for End-to-End Autonomous Drivingcited by 1
  21. 2026MAPLE: Latent Multi-Agent Play for End-to-End Autonomous Drivingcited by 1
  22. 2026World Models as Adversaries: Multi-Agent Self-Play Fine-Tuning for Robust Motion Planningcited by 1
  23. 2026Scaling Curriculum Learning For Autonomous Drivingcited by 1
  24. 2024LASER: Script Execution by Autonomous Agents for On-demand Traffic Simulationcited by 0
  25. 2026Artificial Intelligence for Modeling and Simulation of Mixed Automated and Human Trafficcited by 0
  26. 2026GuardAD: Safeguarding Autonomous Driving MLLMs via Markovian Safety Logiccited by 0
  27. 2026Learning Responsibility-Attributed Adversarial Scenarios for Testing Autonomous Vehiclescited by 0
  28. 2026DriveSafer: End-to-End Autonomous Driving with Safety Guidancecited by 0
  29. 2026KG-ASG: Collision-Knowledge-Guided Closed-Loop Adversarial Scenario Generation With Primary-Support Attributioncited by 0
  30. 2026EvoDrive: Pareto Evolution for Safety-Critical Autonomous Driving via Self-Improving LLM Agentscited by 0
  31. 2026From Attacks to Curricula: Learnability-Guided Adversarial Training for Safe Autonomous Drivingcited by 0
  32. 2026Proposal-Conditioned Latent Diffusion for Closed-Loop Traffic Scenario Generationcited by 0
  33. 2026TrafficAlign: Aligning Large Language Models for Traffic Scenario Generationcited by 0
  34. 2026Agent-driven Long-tail Simulation for Autonomous Drivingcited by 0
  35. 2026CCFM: Collision-Constrained Flow Matching for Safety-Critical Scenario Generationcited by 0
  36. 2026End-to-end Conditional Diffusion for Realistic and Controllable Visual Traffic Scenario Generationcited by 0
  37. 2026Threat-guided Policy-aware Scene Perturbation for Safe Autonomous Driving with Online Reinforcement Learningcited by 0
  38. 2026Language-Structured Relational Q-Learning for Threat-Aware Control in Safety-Critical Drivingcited by 0

50 most cited of 52 All 52 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2024AdvDiffuser: Generating Adversarial Safety-Critical Driving Scenarios via Guided Diffusion
  2. 2026SaFeR: Safety-Critical Scenario Generation for Autonomous Driving Test via Feasibility-Constrained Token Resampling
  3. 2023SAFE-SIM: Safety-Critical Closed-Loop Traffic Simulation with Diffusion-Controllable Adversaries
  4. 2024Adversarial Safety-Critical Scenario Generation using Naturalistic Human Driving Priors
  5. 2025Adversarial Generation and Collaborative Evolution of Safety-Critical Scenarios for Autonomous Vehicles
  6. 2024Module-wise Adaptive Adversarial Training for End-to-end Autonomous Driving
  7. 2024DeepMF: Deep Motion Factorization for Closed-Loop Safety-Critical Driving Scenario Simulation
  8. 2021Adversarial Deep Reinforcement Learning for Improving the Robustness of Multi-agent Autonomous Driving Policies