Tonghe Zhang - Personal Homepage

News

New blog post: Beyond Singularity: The Democratic Promise of Physical AI. | Jun 2026

Released ENPIRE, a coding agent harness for real-world policy self-improvement. | Jun 2026

BFM-zero accepted at ICLR 2026. | Jan 2026

See more ▾

Research

* Equal contribution · ^$ Core contributor

ENPIRE: Agentic Robot Policy Self-Improvement in the Real World

Wenli Xiao*, Jia Xie*, Tonghe Zhang*, Haotian Lin*, Letian "Max" Fu, Haoru Xue, Jalen Lu, Yi Yang, Cunxi Dai, Zi Wang, Jimmy Wu, Guanzhi Wang, S. Shankar Sastry, Ken Goldberg, Linxi "Jim" Fan, Yuke Zhu, Guanya Shi.

Preprint

TL;DR: Coding agents autonomously self-improve real-world dexterous manipulation policies.

Website arXiv Google Scholar

BFM-Zero: A Promptable Behavioral Foundation Model for Humanoid Control Using Unsupervised Reinforcement Learning

Yitang Li*^$, Zhengyi Luo*^$, Tonghe Zhang^$, Cunxi Dai^$, Anssi Kanervisto, Andrea Tirinzoni, Haoyang Weng, Kris Kitani, Mateusz Guzek, Ahmed Touati, Alessandro Lazaric, Matteo Pirotta, Guanya Shi.

ICLR 2025

TL;DR: Humanoid behavior foundation model via unsupervised RL.

Website arXiv GitHub BibTex Google Scholar

@misc{li2025bfmzeropromptablebehavioralfoundation,
      title={BFM-Zero: A Promptable Behavioral Foundation Model for Humanoid Control Using Unsupervised Reinforcement Learning}, 
      author={Yitang Li and Zhengyi Luo and Tonghe Zhang and Cunxi Dai and Anssi Kanervisto and Andrea Tirinzoni and Haoyang Weng and Kris Kitani and Mateusz Guzek and Ahmed Touati and Alessandro Lazaric and Matteo Pirotta and Guanya Shi},
      year={2025},
      eprint={2511.04131},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2511.04131}, 
}

ReinFlow: Fine-tuning Flow Matching Policy with Online Reinforcement Learning

Tonghe Zhang, Chao Yu, Sichang Su, Yu Wang.

NeurIPS 2025

TL;DR: Injecting learnable noise to fine-tune flow matching policies and VLAs with policy gradient.

NeurIPS Homepage Website arXiv Google Scholar GitHub WandB HuggingFace BibTex

@misc{zhang2025reinflowfinetuningflowmatching,
      title={ReinFlow: Fine-tuning Flow Matching Policy with Online Reinforcement Learning}, 
      author={Tonghe Zhang and Chao Yu and Sichang Su and Yu Wang},
      year={2025},
      eprint={2505.22094},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2505.22094}, 
}

π_RL: Online RL Fine-tuning for Flow-based Vision-Language-Action Models

Kang Chen*, Zhihao Liu*, Tonghe Zhang*, Zhen Guo, Si Xu, Hao Lin, Hongzhi Zang, Quanlu Zhang, Zhaofei Yu, Guoliang Fan, Tiejun Huang, Yu Wang, Chao Yu.

Preprint

TL;DR: Online RL fine-tuning for flow-based VLAs (π0, π0.5) with policy gradient, achieving high success rate in LIBERO, ManiSkill, MetaWorld via large-scale parallel simulation.

arXiv GitHub HuggingFace Google Scholar BibTex

@misc{chen2025pitextttrlonlinerlfinetuning,
      title={$\pi_\texttt{RL}$: Online RL Fine-tuning for Flow-based Vision-Language-Action Models},
      author={Kang Chen and Zhihao Liu and Tonghe Zhang and Zhen Guo and Si Xu and Hao Lin and Hongzhi Zang and Quanlu Zhang and Zhaofei Yu and Guoliang Fan and Tiejun Huang and Yu Wang and Chao Yu},
      year={2025},
      eprint={2510.25889},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2510.25889},
}

Sample-Efficient Reinforcement Learning of Flow Policies via Sequence Modeling

Yixian Zhang*, Shu'ang Yu*, Tonghe Zhang, Mo Guang, Haojia Hui, Kaiwen Long, Yu Wang, Chao Yu, Wenbo Ding.

ICLR 2025

TL;DR: Training flow matching policies via off-policy RL by treating the process as RNN/Transformer.

arXiv GitHub Website

Think on Your Feet: Seamless Transition between Human-like Locomotions in Response to Changing Commands

Huaxing Huang*, Wenhao Cui*, Tonghe Zhang*, Shengtao Li, Chenxu Hu, Tianchu Zhang, Liang Zheng, Ziyang Tang, Jinchao Han, Bangyu Qin, Tianchu Zhang, Shipu Zhang, Zheyuan Jiang.

ICRA 2025

TL;DR: Using GAN-based imitation and sim2real RL to teach a humanoid robot to interpolate between various human-like locomotion skills.

IEEE Xplore arXiv Google Scholar BibTex

@INPROCEEDINGS{11127948,
  author={Huang, Huaxing and Cui, Wenhao and Zhang, Tonghe and Li, Shengtao and Han, Jinchao and Qin, Bangyu and Zhang, Tianchu and Zheng, Liang and Tang, Ziyang and Hu, Chenxu and Zhang, Shipu and Jiang, Zheyuan},
  booktitle={2025 IEEE International Conference on Robotics and Automation (ICRA)}, 
  title={Think on Your Feet: Seamless Transition Between Human-Like Locomotion in Response to Changing Commands}, 
  year={2025},
  volume={},
  number={},
  pages={15965-15971},
  keywords={Adaptation models;Tracking;Imitation learning;Stability criteria;Humanoid robots;Multitasking},
  doi={10.1109/ICRA55743.2025.11127948}}

Provably Efficient Partially-observable Risk-sensitive Reinforcement Learning

Tonghe Zhang*, Yu Chen*, Longbo Huang.

ICML 2024

TL;DR: Establishing a statistical theory for the sample complexity of risk-sensitive RL under partial observability.

PMLR arXiv Google Scholar GitHub BibTex

@InProceedings{pmlr-v235-zhang24g,
  title = 	 {Provably Efficient Partially Observable Risk-sensitive Reinforcement Learning with Hindsight Observation},
  author =       {Zhang, Tonghe and Chen, Yu and Huang, Longbo},
  booktitle = 	 {Proceedings of the 41st International Conference on Machine Learning},
  pages = 	 {58680--58716},
  year = 	 {2024},
  editor = 	 {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix},
  volume = 	 {235},
  series = 	 {Proceedings of Machine Learning Research},
  month = 	 {21--27 Jul},
  publisher =    {PMLR},
  pdf = 	 {https://raw.githubusercontent.com/mlresearch/v235/main/assets/zhang24g/zhang24g.pdf},
  url = 	 {https://proceedings.mlr.press/v235/zhang24g.html},
}