TeViR: Text-to-Video Reward with Diffusion Models for Efficient Reinforcement Learning

2025-05-26 · Yuhui Chen, Haoran Li, Zhennan Jiang, Haowei Wen, Dongbin Zhao

Developing scalable and generalizable reward engineering for reinforcement learning (RL) is crucial for creating general-purpose agents, especially in the challenging domain of robotic manipulation. While recent advances in reward engineering with Vision-Language Models (VLMs) have shown promise, their sparse reward nature significantly limits sample efficiency. This paper introduces TeViR, a novel method that leverages a pre-trained text-to-video diffusion model to generate dense rewards by comparing the predicted image sequence with current observations. Experimental results across 11 complex robotic tasks demonstrate that TeViR outperforms traditional methods leveraging sparse rewards and other state-of-the-art (SOTA) methods, achieving better sample efficiency and performance without ground truth environmental rewards. TeViR's ability to efficiently guide agents in complex environments highlights its potential to advance reinforcement learning applications in robotic manipulation.

📄 PDF Abstract BibTeX arXiv:2505.19769

Code (0)

등록된 구현이 없습니다.

Tasks

reinforcement-learning Reinforcement Learning Reinforcement Learning (RL)

Methods 이 논문이 사용한 방법론

Diffusion Diffusion models generate samples by gradually removing noise from a signal, and their training objective can be expressed as a reweighted variational lower-bound…

Diffusion Reward: Learning Rewards via Conditional Video Diffusion

2023-12-21 · Tao Huang, Guangqi Jiang, Yanjie Ze, Huazhe Xu

Learning rewards from expert videos offers an affordable and effective solution to specify the intended behaviors for reinforcement learning (RL) tasks. In this work, we propose Diffusion Reward, a novel framework that l…

Diversity Reinforcement Learning (RL)

Goal-Driven Reward by Video Diffusion Models for Reinforcement Learning

2025-11-30 · Qi Wang, Mian Wu, Yuyang Zhang, Mingqi Yuan 외 arxiv

Reinforcement Learning (RL) has achieved remarkable success in various domains, yet it often relies on carefully designed programmatic reward functions to guide agent behavior. Designing such reward functions can be chal…

Reinforcement Learning

DanceGRPO: Unleashing GRPO on Visual Generation

2025-05-12 · Zeyue Xue, Jie Wu, Yu Gao, Fangyuan Kong 외

Recent breakthroughs in generative models-particularly diffusion models and rectified flows-have revolutionized visual content creation, yet aligning model outputs with human preferences remains a critical challenge. Exi…

Denoising reinforcement-learning Reinforcement Learning Reinforcement Learning (RL)+1

Free$^2$Guide: Gradient-Free Path Integral Control for Enhancing Text-to-Video Generation with Large Vision-Language Models

2024-11-26 · JaeMin Kim, Bryan S Kim, Jong Chul Ye

Diffusion models have achieved impressive results in generative tasks like text-to-image (T2I) and text-to-video (T2V) synthesis. However, achieving accurate text alignment in T2V generation remains challenging due to th…

Reinforcement Learning (RL)Text-to-Video Generation Video Generation

Data-regularized Reinforcement Learning for Diffusion Models at Scale

2025-12-03 · Haotian Ye, Kaiwen Zheng, Jiashu Xu, Puheng Li 외 arxiv

Aligning generative diffusion models with human preferences via reinforcement learning (RL) is critical yet challenging. Most existing algorithms are often vulnerable to reward hacking, such as quality degradation, over-…

Reinforcement Learning Video Generation