mlsanityData to 14 Oct 2026

TrafficAlign: Aligning Large Language Models for Traffic Scenario Generation

arXiv 2606.29097 ↗First posted 27 Jun 2026cs.CVcited by 0

On the map: LLM alignment / RLHF / preference optimisation → Reward modeling and RLHF

Abstract (arXiv)

Recent research has investigated the use of large language models (LLMs) to generate traffic scenarios for autonomous driving. However, pretrained LLMs often fail to align with real-world traffic distributions. In this work, we present TrafficAlign, an automated framework that synthesizes traffic scenarios based on real-world driving videos, performs data validation, and aligns LLMs with the synthesized scenarios. The evaluation shows that traffic scenarios generated by TrafficAlign are highly effective, revealing up to 10.8% more collisions on average across three autonomous driving models than state-of-the-art methods. Furthermore, fine-tuning these driving models with TrafficAlign-generated scenarios significantly reduced collision rates by 36.1% compared with the original models. A qualitative study using traffic datasets from six geographically diverse regions shows that TrafficAlign-generated scenarios exhibit strong alignment with corresponding traffic distributions in these regions.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Reward modeling and RLHF, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 45 of its 69 references that are in the map. First posted on arXiv 27 Jun 2026.

Builds on

45 of 69 references.

Closest ancestors · 1

Same direction, newest first

  1. 2024Self-Rewarding Language Models18 Jan

Other references · 44

  1. 2020Language Models are Few-Shot Learnerscited by 32,419
  2. 2017Proximal Policy Optimization Algorithmscited by 15,215
  3. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
  4. 2019Sentence-BERT: Sentence Embeddings using Siamese BERT-Networkscited by 9,170
  5. 2025Qwen3 Technical Reportcited by 8,820
  6. 2018Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actorcited by 4,885
  7. 2024GPT-4o System Cardcited by 4,588
  8. 2018UMAP: Uniform Manifold Approximation and Projection for Dimension Reductioncited by 2,898
  9. 2022Self-Instruct: Aligning Language Models with Self-Generated Instructionscited by 2,724
  10. 2017CARLA: An Open Urban Driving Simulatorcited by 2,449
  11. 2018Addressing Function Approximation Error in Actor-Critic Methodscited by 2,417
  12. 2023WizardLM: Empowering large pre-trained language models to follow complex instructionscited by 1,125
32 more
  1. 2023Role-Play with Large Language Modelscited by 294
  2. 2021TrafficSim: Learning to Simulate Realistic Multi-Agent Behaviorscited by 193
  3. 2024The PRISM Alignment Dataset: What Participatory, Representative and Individualised Human Feedback Reveals About the Subjective and Multicultural Alignment of Large Language Modelscited by 144
  4. 2021Generating Useful Accident-Prone Driving Scenarios via a Learned Traffic Priorcited by 139
  5. 2019Meta-Sim: Learning to Generate Synthetic Datasetscited by 128
  6. 2024Collective Constitutional AI: Aligning a Language Model with Public Inputcited by 123
  7. 2024ChatScene: Knowledge-Enabled Safety-Critical Scenario Generation for Autonomous Vehiclescited by 93
  8. 2022On Adversarial Robustness of Trajectory Prediction for Autonomous Vehiclescited by 90
  9. 2023Language Conditioned Traffic Generationcited by 87
  10. 2020Learning to Collide: An Adaptive Safety-Critical Scenarios Generating Methodcited by 75
  11. 2020SurfelGAN: Synthesizing Realistic Sensor Data for Autonomous Drivingcited by 68
  12. 2022SafeBench: A Benchmarking Platform for Safety Evaluation of Autonomous Vehiclescited by 66
  13. 2024SelfCodeAlign: Self-Alignment for Code Generationcited by 62
  14. 2020Meta-Sim2: Unsupervised Learning of Scene Structure for Synthetic Data Generationcited by 58
  15. 2023CAT: Closed-loop Adversarial Training for Safe End-to-End Drivingcited by 54
  16. 2023Scenario Diffusion: Controllable Driving Scenario Generation With Diffusioncited by 47
  17. 2024FLAME: Factuality-Aware Alignment for Large Language Modelscited by 47
  18. 2017Ontology based Scene Creation for the Development of Automated Vehiclescited by 40
  19. 2020Scenic: A Language for Scenario Specification and Data Generationcited by 40
  20. 2024CodecLM: Aligning Language Models with Tailored Synthetic Datacited by 39
  21. 2024SLEDGE: Synthesizing Driving Environments with Generative Models and Rule-Based Trafficcited by 27
  22. 2019CMTS: Conditional Multiple Trajectory Synthesizer for Generating Safety-critical Driving Scenarioscited by 19
  23. 2022DriveFuzz: Discovering Autonomous Driving Bugs through Driving Quality-Guided Fuzzingcited by 19
  24. 2022LawBreaker: An Approach for Specifying Traffic Laws and Fuzzing Autonomous Vehiclescited by 16
  25. 2023BehAVExplor: Behavior Diversity Guided Testing for Autonomous Driving Systemscited by 11
  26. 2024ScenicNL: Generating Probabilistic Scenario Programs from Natural Languagecited by 10
  27. 2023TARGET: Automated Scenario Generation from Traffic Rules for Testing Autonomous Vehicles via Validated LLM-Guided Knowledge Extractioncited by 9
  28. 2024SoVAR: Building Generalizable Scenarios from Accident Reports for Autonomous Driving Testingcited by 9
  29. 2024LeGEND: A Top-Down Approach to Scenario Generation of Autonomous Driving Systems Assisted by Large Language Modelscited by 9
  30. 2025Adversarial Preference Learning for Robust LLM Alignmentcited by 7
  31. 2024Generating Traffic Scenarios via In-Context Learning to Learn Better Motion Plannercited by 6
  32. 2025Multi-modal Traffic Scenario Generation for Autonomous Driving System Testingcited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2024Large Language Models Powered Context-aware Motion Prediction in Autonomous Driving
  2. 2025DriveGen: Towards Infinite Diverse Traffic Scenarios with Large Models
  3. 2026Long-term Traffic Simulation via Structured Autoregressive Modeling
  4. 2023ScenarioNet: Open-Source Platform for Large-Scale Traffic Scenario Simulation and Modeling
  5. 2024Driving Everywhere with Large Language Model Policy Adaptation
  6. 2024Traffic Scene Generation from Natural Language Description for Autonomous Vehicles with Large Language Model
  7. 2025LLM-based Realistic Safety-Critical Driving Video Generation
  8. 2025A Tool for Benchmarking Large Language Models' Robustness in Assessing the Realism of Driving Scenarios ↗