Sequential decision-making
Learn representations that improve action. We study representation learning, model-based planning, and reward construction for agents that must learn from pixels, sparse feedback, and changing environments.
Research question
What should an agent remember, predict, and explore in order to act well over time?
We study representation learning, model-based planning, and reward construction for agents that must learn from pixels, sparse feedback, and changing environments.
- Visual reinforcement learning
- Model-based planning
- Object-centric rewards
- Representation learning
Representative projects
Learn representations that improve action.
Each project connects its central idea with papers, code, datasets, demonstrations, and public explanations.
Selected publications
The papers behind the projects.
Browse the full publication database for the broader body of work in world models.
GenFlowRL: Shaping Rewards with Generative Object-Centric Flow in Visual Reinforcement Learning
International Conference on Computer Vision (ICCV), 2025
BibTeX
@inproceedings{yu2025genflowrl0028,
title = {GenFlowRL: Shaping Rewards with Generative Object-Centric Flow in Visual Reinforcement Learning},
author = {Kelin Yu and Sheng Zhang and Harshit Soora and Furong Huang and Heng Huang and Pratap Tokekar and Ruohan Gao},
booktitle = {International Conference on Computer Vision (ICCV), 2025},
year = {2025},
eprint = {2508.11049},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2508.11049},
}DrM: Mastering Visual Reinforcement Learning through Dormant Ratio Minimization
Spotlight. The Twelfth International Conference on Learning Representations (ICLR), 2024
BibTeX
@inproceedings{xu2024mastering32d0,
title = {DrM: Mastering Visual Reinforcement Learning through Dormant Ratio Minimization},
author = {Guowei Xu and Ruijie Zheng and Yongyuan Liang and Xiyao Wang and Zhecheng Yuan and Tianying Ji and Yu Luo and Xiaoyu Liu and Jiaxin Yuan and Pu Hua and Shuzhen Li and Yanjie Ze and Hal Daume III and Furong Huang and Huazhe Xu},
booktitle = {Spotlight. The Twelfth International Conference on Learning Representations (ICLR), 2024},
year = {2024},
eprint = {2310.19668},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2310.19668},
}TACO: Temporal Latent Action-Driven Contrastive Loss for Visual Reinforcement Learning
The Thirty-seventh Annual Conference on Neural Information Processing Systems (NeurIPS), 2023
BibTeX
@inproceedings{zheng2023tacod045,
title = {TACO: Temporal Latent Action-Driven Contrastive Loss for Visual Reinforcement Learning},
author = {Ruijie Zheng and Xiyao Wang and Yanchao Sun and Shuang Ma and Jieyu Zhao and Huazhe Xu and Hal Daume III and Furong Huang},
booktitle = {The Thirty-seventh Annual Conference on Neural Information Processing Systems (NeurIPS), 2023},
year = {2023},
eprint = {2306.13229},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2306.13229},
}COPlanner: Plan to Roll Out Conservatively but to Explore Optimistically for Model-Based RL
The Twelfth International Conference on Learning Representations (ICLR), 2024
BibTeX
@inproceedings{wang2024coplanner3f46,
title = {COPlanner: Plan to Roll Out Conservatively but to Explore Optimistically for Model-Based RL},
author = {Xiyao Wang and Ruijie Zheng and Yanchao Sun and Ruonan Jia and Wichayaporn Wongkamjan and Huazhe Xu and Furong Huang},
booktitle = {The Twelfth International Conference on Learning Representations (ICLR), 2024},
year = {2024},
eprint = {2310.07220},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2310.07220},
}