Email Google Scholar GitHub

HuggingFace LinkedIn X (formerly Twitter)

Follow @TongheZhang01

Tonghe Zhang

I am a Master of Science student in Robotics at Robotics Institute, Carnegie Mellon University.

I study how to build self-improving robot systems that act efficiently and reliably in the real world.

Email: tonghez[AT]andrew.cmu.edu

News

[2025-06-26] ReinFlow is now online! This work demonstrates how to fine-tune flow policies with reinforcement learning. Check it out on arXiv, HuggingFace, and GitHub

Talks

[2025-06-16] Delivered a talk at Tsinghua University on “Designing Practical Robot Learning Pipelines with Flow Policies and Reinforcement Learning.”

Research

* denotes equal contribution

Preprints

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

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

Preprint, under review.

TL;DR: Injecting learnable stochastic noise to post-train flow matching policies (e.g., 1-Rectified Flow and Shortcut Models) with policy gradient.

Website arXiv Google Scholar GitHub 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}, 
}

Publications

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.

2025 IEEE International Conference on Robotics and Automation (ICRA 2025).

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

arXiv Google Scholar BibTex

@misc{huang2025thinkfeetseamlesstransition,
      title={Think on your feet: Seamless Transition between Human-like Locomotion in Response to Changing Commands}, 
      author={Huaxing Huang and Wenhao Cui and Tonghe Zhang and Shengtao Li and Jinchao Han and Bangyu Qin and Tianchu Zhang and Liang Zheng and Ziyang Tang and Chenxu Hu and Ning Yan and Jiahao Chen and Shipu Zhang and Zheyuan Jiang},
      year={2025},
      eprint={2502.18901},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2502.18901}, 
}

Provably Efficient Partially-observable Risk-sensitive Reinforcement Learning

Tonghe Zhang*, Yu Chen*, Longbo Huang.

Proceedings of the 41st International Conference on Machine Learning (ICML 2024).

TL;DR: Establishing a statistical theory for 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},
}