paper-with-me

홈 › Papers

Pseudo-Expert Regularized Offline RL for End-to-End Autonomous Driving in Photorealistic Closed-Loop Environments

2025-12-21 · Chihiro Noguchi, Takaki Yamamoto arxiv

End-to-end (E2E) autonomous driving models that take only camera images as input and directly predict a future trajectory are appealing for their computational efficiency and potential for improved generalization via unified optimization; however, persistent failure modes remain due to reliance on imitation learning (IL). While online reinforcement learning (RL) could mitigate IL-induced issues, the computational burden of neural rendering-based simulation and large E2E networks renders iterative reward and hyperparameter tuning costly. We introduce a camera-only E2E offline RL framework that performs no additional exploration and trains solely on a fixed simulator dataset. Offline RL offers strong data efficiency and rapid experimental iteration, yet is susceptible to instability from overestimation on out-of-distribution (OOD) actions. To address this, we construct pseudo ground-truth trajectories from expert driving logs and use them as a behavior regularization signal, suppressing imitation of unsafe or suboptimal behavior while stabilizing value learning. Training and closed-loop evaluation are conducted in a neural rendering environment learned from the public nuScenes dataset. Empirically, the proposed method achieves substantial improvements in collision rate and route completion compared with IL baselines. Our code is available at https://github.com/ToyotaInfoTech/PEBC.

📄 PDF Abstract BibTeX arXiv:2512.18662

Code (0)

등록된 구현이 없습니다.

Tasks

Computational EfficiencyReinforcement LearningAutonomous DrivingOffline RL

Similar Papers 제목 키워드 기반

(Re)$^2$H2O: Autonomous Driving Scenario Generation via Reversely Regularized Hybrid Offline-and-Online Reinforcement Learning

2023-02-27 · Haoyi Niu, Kun Ren, Yizhou Xu, Ziyuan Yang 외

Autonomous driving and its widespread adoption have long held tremendous promise. Nevertheless, without a trustworthy and thorough testing procedure, not only does the industry struggle to mass-produce autonomous vehicle…

Autonomous DrivingAutonomous Vehicles

CoIRL-AD: Collaborative-Competitive Imitation-Reinforcement Learning in Latent World Models for Autonomous Driving

2025-10-14 · Xiaoji Zheng, Ziyuan Yang, Yanhao Chen, Yuhang Peng 외 arxiv

End-to-end autonomous driving models trained with imitation learning (IL) often generalize poorly, particularly in long-tail scenarios where expert demonstrations are sparse. Reinforcement learning (RL) can provide compl…

Reinforcement LearningAutonomous Driving

PerlAD: Towards Enhanced Closed-loop End-to-end Autonomous Driving with Pseudo-simulation-based Reinforcement Learning

2026-03-16 · Yinfeng Gao, Qichao Zhang, Deqing Liu, Zhongpu Xia 외 arxiv

End-to-end autonomous driving policies based on Imitation Learning (IL) often struggle in closed-loop execution due to the misalignment between inadequate open-loop training objectives and real driving requirements. Whil…

Reinforcement LearningAutonomous Driving

Guided Online Distillation: Promoting Safe Reinforcement Learning by Offline Demonstration

2023-09-18 · Jinning Li, Xinyi Liu, Banghua Zhu, Jiantao Jiao 외

Safe Reinforcement Learning (RL) aims to find a policy that achieves high rewards while satisfying cost constraints. When learning from scratch, safe RL agents tend to be overly conservative, which impedes exploration an…

Autonomous DrivingDecision Makingreinforcement-learningReinforcement Learning+2

GRI: General Reinforced Imitation and its Application to Vision-Based Autonomous Driving

2021-11-16 · Raphael Chekroun, Marin Toromanoff, Sascha Hornauer, Fabien Moutarde

Deep reinforcement learning (DRL) has been demonstrated to be effective for several complex decision-making applications such as autonomous driving and robotics. However, DRL is notoriously limited by its high sample com…

Autonomous DrivingCARLA MAP Leaderboardcontinuous-controlContinuous Control+3