mlsanityData to 14 Oct 2026

Learning Safe Autonomous Driving Policies Using Predictive Safety Representations

arXiv 2512.17586 ↗First posted 19 Dec 2025cs.LG cs.ROcited by 0

On the map: End-to-end autonomous driving → Reinforcement learning for driving

also in 1 more
Abstract (arXiv)

Safe reinforcement learning (SafeRL) is a prominent paradigm for autonomous driving, where agents are required to optimize performance under strict safety requirements. This dual objective creates a fundamental tension, as overly conservative policies limit driving efficiency while aggressive exploration risks safety violations. The Safety Representations for Safer Policy Learning (SRPL) framework addresses this challenge by equipping agents with a predictive model of future constraint violations and has shown promise in controlled environments. This paper investigates whether SRPL extends to real-world autonomous driving scenarios. Systematic experiments on the Waymo Open Motion Dataset (WOMD) and NuPlan demonstrate that SRPL can improve the reward-safety tradeoff, achieving statistically significant improvements in success rate (effect sizes r = 0.65-0.86) and cost reduction (effect sizes r = 0.70-0.83), with p < 0.05 for observed improvements. However, its effectiveness depends on the underlying policy optimizer and the dataset distribution. The results further show that predictive safety representations play a critical role in improving robustness to observation noise. Additionally, in zero-shot cross-dataset evaluation, SRPL-augmented agents demonstrate improved generalization compared to non-SRPL methods. These findings collectively demonstrate the potential of predictive safety representations to strengthen SafeRL for autonomous driving.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Reinforcement learning for driving, one of 9 directions in End-to-end autonomous driving. Builds on 26 of its 33 references that are in the map. First posted on arXiv 19 Dec 2025.

Builds on

26 of 33 references.

Closest ancestors · 1

Same direction, newest first

  1. 2021MetaDrive: Composing Diverse Driving Scenarios for Generalizable Reinforcement Learning26 Sep

Other references · 25

  1. 2018Addressing Function Approximation Error in Actor-Critic Methodscited by 2,417
  2. 2018World Modelscited by 1,513
  3. 2017Constrained Policy Optimizationcited by 983
  4. 2021Deep Reinforcement Learning at the Edge of the Statistical Precipicecited by 807
  5. 2020Deep Reinforcement Learning for Autonomous Driving: A Surveycited by 721
  6. 2021Large Scale Interactive Motion Forecasting for Autonomous Driving : The Waymo Open Motion Datasetcited by 537
  7. 2021NuPlan: A closed-loop ML-based planning benchmark for autonomous vehiclescited by 493
  8. 2016Reinforcement Learning with Unsupervised Auxiliary Taskscited by 404
  9. 2016Benchmarking Deep Reinforcement Learning for Continuous Controlcited by 371
  10. 2020Data-Efficient Reinforcement Learning with Self-Predictive Representationscited by 318
  11. 2020Responsive Safety in Reinforcement Learning by PID Lagrangian Methodscited by 316
  12. 2022Motion Transformer with Global Intention Localization and Local Movement Refinementcited by 270
13 more
  1. 2023MotionLM: Multi-Agent Motion Forecasting as Language Modelingcited by 156
  2. 2020CRPO: A New Approach for Safe Reinforcement Learning with Convergence Guaranteecited by 134
  3. 2022Imitation Is Not Enough: Robustifying Imitation with Reinforcement Learning for Challenging Driving Scenarioscited by 129
  4. 2023OmniSafe: An Infrastructure for Accelerating Safe Reinforcement Learning Researchcited by 80
  5. 2022Penalized Proximal Policy Optimization for Safe Reinforcement Learningcited by 68
  6. 2023ScenarioNet: Open-Source Platform for Large-Scale Traffic Scenario Simulation and Modelingcited by 60
  7. 2020Can Increasing Input Dimensionality Improve Deep Reinforcement Learning?cited by 37
  8. 2023Safe Exploration in Reinforcement Learning: A Generalized Formulation and Algorithmscited by 28
  9. 2022When to Ask for Help: Proactive Interventions in Autonomous Reinforcement Learningcited by 21
  10. 2024CIMRL: Combining IMitation and Reinforcement Learning for Safe Autonomous Drivingcited by 12
  11. 2025Safety Representations for Safer Policy Learningcited by 4
  12. 2023Graph-based Prediction and Planning Policy Network (GP3Net) for scalable self-driving in dynamic environments using Deep Reinforcement Learningcited by 3
  13. 2024Deep Attention Driven Reinforcement Learning (DAD-RL) for Autonomous Decision-Making in Dynamic Environmentcited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Safety Reinforced Model Predictive Control (SRMPC): Improving MPC with Reinforcement Learning for Motion Planning in Autonomous Driving
  2. 2025Safe Reinforcement Learning with a Predictive Safety Filter for Motion Planning and Control: A Drifting Vehicle Example
  3. 2022SafeRL-Kit: Evaluating Efficient Reinforcement Learning Methods for Safe Autonomous Driving
  4. 2022How to Learn from Risk: Explicit Risk-Utility Reinforcement Learning for Efficient and Safe Driving Strategies
  5. 2025Balancing Progress and Safety: A Novel Risk-Aware Objective for RL in Autonomous Driving
  6. 2021Minimizing Safety Interference for Safe and Comfortable Automated Driving with Distributional Reinforcement Learning
  7. 2024Latent Safety-Constrained Policy Approach for Safe Offline Reinforcement Learning
  8. 2024Long and Short-Term Constraints Driven Safe Reinforcement Learning for Autonomous Driving