End-to-End Model-Free Reinforcement Learning for Urban Driving using\n Implicit Affordances

2019/11/25 by Marin Toromanoff, Toromanoff, Marin, Émilie Wirbel +3 · 16 citations
Computer Science · Engineering · #Artificial Intelligence (cs.AI) #Autonomous Vehicle Technology and Safety #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Reinforcement Learning in Robotics #Robotics (cs.RO) #Traffic control and management

paper · pdf · doi:10.48550/arxiv.1911.10868

openalex publication_date 2019/11/25 · openalex created_date 2022/12/28 · openalex updated_date 2026/07/28

Abstract

Reinforcement Learning (RL) aims at learning an optimal behavior policy from\nits own experiments and not rule-based control methods. However, there is no RL\nalgorithm yet capable of handling a task as difficult as urban driving. We\npresent a novel technique, coined implicit affordances, to effectively leverage\nRL for urban driving thus including lane keeping, pedestrians and vehicles\navoidance, and traffic light detection. To our knowledge we are the first to\npresent a successful RL agent handling such a complex task especially regarding\nthe traffic light detection. Furthermore, we have demonstrated the\neffectiveness of our method by winning the Camera Only track of the CARLA\nchallenge.\n

Cited by

• InDRiVE: Reward-Free World-Model Pretraining for Autonomous Driving via Latent Disagreement
• Pseudo-Expert Regularized Offline RL for End-to-End Autonomous Driving in Photorealistic Closed-Loop Environments
• Vision-Language-Action Models for Autonomous Driving: Past, Present, and Future
• FutureX: Enhance End-to-End Autonomous Driving via Latent Chain-of-Thought World Model
• WPT: World-to-Policy Transfer via Online World Model Distillation
• Attacking Autonomous Driving Agents with Adversarial Machine Learning: A Holistic Evaluation with the CARLA Leaderboard
• Embodied Cognition Augmented End2End Autonomous Driving
• AdaDrive: Self-Adaptive Slow-Fast System for Language-Grounded Autonomous Driving
• VLDrive: Vision-Augmented Lightweight MLLMs for Efficient Language-grounded Autonomous Driving
• ZTRS: Zero-Imitation End-to-end Autonomous Driving with Trajectory Scoring
• VR-Drive: Viewpoint-Robust End-to-End Driving with Feed-Forward 3D Gaussian Splatting
• RAP: 3D Rasterization Augmented End-to-End Planning
• OmniScene: Attention-Augmented Multimodal 4D Scene Understanding for Autonomous Driving
• DriveDPO: Policy Learning via Safety DPO For End-to-End Autonomous Driving
• Temporal Logic-Based Multi-Vehicle Backdoor Attacks against Offline RL Agents in End-to-end Autonomous Driving
• Interpretable Decision-Making for End-to-End Autonomous Driving

Related

• Affordance-based Reinforcement Learning for Urban Driving
• CADRE: A Cascade Deep Reinforcement Learning Framework for Vision-based Autonomous Urban Driving
• Conditional Affordance Learning for Driving in Urban Environments
• Privileged to Predicted: Towards Sensorimotor Reinforcement Learning for Urban Driving
• Model-free Deep Reinforcement Learning for Urban Autonomous Driving