Rui Yang

Rui Yang

Hi, I am Rui Yang, a senior undergraduate student from the School of Computer Science, Peking University (PKU). At Peking University, I worked with Prof. Hao Tang at the PKU TH-AI Lab. I spent a summer working with Prof. Haimin Hu at Johns Hopkins University (JHU), where I joined the Laboratory for Computational Sensing & Robotics (LCSR) and the Alliance AI Lab. My research interests broadly lie in Robotics, Safe Reinforcement Learning, and Computer Vision.

Publications

arXiv

Turning Safety into Competence: Minimally Exploitable Robot Policies via Safety-Filtered Reinforcement Learning

Ruihan Wu*, Rui Yang*, Donggeon David Oh, Duy P. Nguyen, and Haimin Hu

arXiv preprint arXiv:2609.27312, 2026  ·  *Equal contribution

arXiv HTML PDF Website Code

We propose Safety to Competence (S2C), a two-stage reinforcement learning framework that separates safety synthesis from competitive task learning. S2C learns a robust safety filter through adversarial RL, embeds it during task-policy training, and retains it at deployment. In simulated touchdown games and hardware stress tests against a human opponent, S2C improves competitive performance and reduces exploitability while maintaining safety.

ECCV 2026

MobileVLA-R1: Reinforcing Vision-Language-Action for Mobile Robots

Ting Huang*, Dongjian Li*, Rui Yang*, Zeyu Zhang*‡, Zida Yang, and Hao Tang‡

European Conference on Computer Vision (ECCV), pp. 341–361, 2026  ·  *Equal contribution

arXiv HTML PDF Website Code

MobileVLA-R1 is a unified vision-language-action framework that connects natural-language reasoning with continuous quadruped control. It combines multi-granularity chain-of-thought supervision with GRPO reinforcement learning to improve reasoning consistency, control stability, and long-horizon execution. Evaluations on VLN and VLA tasks, together with real-world deployment, demonstrate robust performance in complex environments.

BibTeX

@article{s2c2026,
  title   = {Turning Safety into Competence: Minimally Exploitable Robot Policies via Safety-Filtered Reinforcement Learning},
  author  = {Wu, Ruihan and Yang, Rui and Oh, Donggeon David and Nguyen, Duy P. and Hu, Haimin},
  journal = {arXiv preprint arXiv:2609.27312},
  year    = {2026}
}

BibTeX

@inproceedings{huang2026mobilevla,
  title     = {MobileVLA-R1: Reinforcing Vision-Language-Action for Mobile Robots},
  author    = {Huang, Ting and Li, Dongjian and Yang, Rui and Zhang, Zeyu and Yang, Zida and Tang, Hao},
  booktitle = {European Conference on Computer Vision},
  pages     = {341--361},
  year      = {2026}
}