ReconDreamer-RL: Enhancing Reinforcement Learning via Diffusion-based Scene Reconstruction

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ni, Chaojun, Zhao, Guosheng, Wang, Xiaofeng, Zhu, Zheng, Qin, Wenkang, Chen, Xinze, Jia, Guanghong, Huang, Guan, Mei, Wenjun
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916910469218304
author Ni, Chaojun
Zhao, Guosheng
Wang, Xiaofeng
Zhu, Zheng
Qin, Wenkang
Chen, Xinze
Jia, Guanghong
Huang, Guan
Mei, Wenjun
author_facet Ni, Chaojun
Zhao, Guosheng
Wang, Xiaofeng
Zhu, Zheng
Qin, Wenkang
Chen, Xinze
Jia, Guanghong
Huang, Guan
Mei, Wenjun
contents Reinforcement learning for training end-to-end autonomous driving models in closed-loop simulations is gaining growing attention. However, most simulation environments differ significantly from real-world conditions, creating a substantial simulation-to-reality (sim2real) gap. To bridge this gap, some approaches utilize scene reconstruction techniques to create photorealistic environments as a simulator. While this improves realistic sensor simulation, these methods are inherently constrained by the distribution of the training data, making it difficult to render high-quality sensor data for novel trajectories or corner case scenarios. Therefore, we propose ReconDreamer-RL, a framework designed to integrate video diffusion priors into scene reconstruction to aid reinforcement learning, thereby enhancing end-to-end autonomous driving training. Specifically, in ReconDreamer-RL, we introduce ReconSimulator, which combines the video diffusion prior for appearance modeling and incorporates a kinematic model for physical modeling, thereby reconstructing driving scenarios from real-world data. This narrows the sim2real gap for closed-loop evaluation and reinforcement learning. To cover more corner-case scenarios, we introduce the Dynamic Adversary Agent (DAA), which adjusts the trajectories of surrounding vehicles relative to the ego vehicle, autonomously generating corner-case traffic scenarios (e.g., cut-in). Finally, the Cousin Trajectory Generator (CTG) is proposed to address the issue of training data distribution, which is often biased toward simple straight-line movements. Experiments show that ReconDreamer-RL improves end-to-end autonomous driving training, outperforming imitation learning methods with a 5x reduction in the Collision Ratio.
format Preprint
id arxiv_https___arxiv_org_abs_2508_08170
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle ReconDreamer-RL: Enhancing Reinforcement Learning via Diffusion-based Scene Reconstruction
Ni, Chaojun
Zhao, Guosheng
Wang, Xiaofeng
Zhu, Zheng
Qin, Wenkang
Chen, Xinze
Jia, Guanghong
Huang, Guan
Mei, Wenjun
Computer Vision and Pattern Recognition
Reinforcement learning for training end-to-end autonomous driving models in closed-loop simulations is gaining growing attention. However, most simulation environments differ significantly from real-world conditions, creating a substantial simulation-to-reality (sim2real) gap. To bridge this gap, some approaches utilize scene reconstruction techniques to create photorealistic environments as a simulator. While this improves realistic sensor simulation, these methods are inherently constrained by the distribution of the training data, making it difficult to render high-quality sensor data for novel trajectories or corner case scenarios. Therefore, we propose ReconDreamer-RL, a framework designed to integrate video diffusion priors into scene reconstruction to aid reinforcement learning, thereby enhancing end-to-end autonomous driving training. Specifically, in ReconDreamer-RL, we introduce ReconSimulator, which combines the video diffusion prior for appearance modeling and incorporates a kinematic model for physical modeling, thereby reconstructing driving scenarios from real-world data. This narrows the sim2real gap for closed-loop evaluation and reinforcement learning. To cover more corner-case scenarios, we introduce the Dynamic Adversary Agent (DAA), which adjusts the trajectories of surrounding vehicles relative to the ego vehicle, autonomously generating corner-case traffic scenarios (e.g., cut-in). Finally, the Cousin Trajectory Generator (CTG) is proposed to address the issue of training data distribution, which is often biased toward simple straight-line movements. Experiments show that ReconDreamer-RL improves end-to-end autonomous driving training, outperforming imitation learning methods with a 5x reduction in the Collision Ratio.
title ReconDreamer-RL: Enhancing Reinforcement Learning via Diffusion-based Scene Reconstruction
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2508.08170