Zhang Yuyou

Yuyou Zhang

Home

CV


Contact:
yuyouz AT andrew DOT cmu DOT edu

Hosted on GitHub Pages — Theme by orderedlist

About Me

I am a fourth-year Ph.D. candidate at Safe AI Lab at CMU advised by Prof. Ding Zhao. I worked at IRMV Lab advised by Prof. Hesheng Wang and got my bachelor degree from Shanghai Jiao Tong University in 2022. I also worked with MSC Lab at UC Berkeley in 2021 advised by Prof. Masayoshi Tomizuka.

My research interests lie at the intersection of embodied AI, reinforcement learning, and multimodal foundation models. I am particularly focused on developing intelligent agents capable of learning and adapting in dynamic, real-world environments. This includes improving decision-making, control, and safety within embodied systems.

News and Updates

  • Sep 2025: We introduce SpinBench, a diagnostic benchmark for spatial reasoning in VLMs. Dataset and code are released! Check our website for more details.
  • Jul 2025: Bipedalism for Quadrupedal Robots accepted to Humanoids, check out our videos here.
  • Jun 2025: QuietPaw accepted to IROS
  • May 2025: Two papers accepted to ACL
  • May 2024: I am starting an internship with the Optimization & Intelligent Robotics team at MERL!
  • Jan 2024: One paper accepted to ICRA
  • Sep 2023: I passed the qualification and became a PhD candidate
  • Sep 2022: One paper accepted to IEEE T-Mech
  • June 2022: I received Zhiyuan Honors Degree and was awarded Shanghai Outstanding Graduate
  • May 2022: One paper accepted to IEEE T-CST

    • Publcations

    hpp

    SpinBench: Perspective and Rotation as a Lens on Spatial Reasoning in VLMs
    Y. Zhang, R. Corcodel, C. Hori, A. Cherian, D. Zhao
    Under review
    PDF, Website, Dataset, Code

    hpp

    Bipedalism for Quadrupedal Robots: Versatile Loco-Manipulation through Risk-Adaptive Reinforcement Learning
    Y. Zhang, R. Corcodel, and D. Zhao
    IEEE International Conference on Humanoid Robots, 2025
    PDF, video

    hpp

    Safety is Not Only About Refusal: Reasoning-Enhanced Fine-tuning for Interpretable LLM Safety
    Y. Zhang, M. Li, W. Han, Y. Yao, Z. Cen, and D. Zhao
    ACL 2025 Findings
    PDF

    hpp

    QuietPaw: Learning Quadrupedal Locomotion with Versatile Noise Preference Alignment
    Y. Zhang, Y. Yao, S. Liu, Y. Niu, C. Lin, Y. Yang, W. Yu, T. Zhang, J. Tan, and D. Zhao
    International Conference on Intelligent Robots and Systems (IROS), 2025
    PDF

    hpp

    COMPOSER: Scalable and Robust Modular Policies for Snake Robots
    Y. Zhang, Y. Niu, X. Liu, and D. Zhao
    IEEE International Conference on Robotics and Automation (ICRA), 2024
    Abridged in CoRL 2023 Workshop on Learning for Soft Robots: Hard Challenges for Soft Systems (Spotlight)
    PDF, website

    hpp

    Offline-Online Learning of Deformation Model for Cable Manipulation with Graph Neural Networks
    C. Wang, Y. Zhang, X. Zhang, Z. Wu, X. Zhu, S. Jin, T. Tang, and M. Tomizuka
    IEEE Robotics and Automation Letters (RA-L), 2022
    PDF, website

    hpp

    Vision-Based Contact Point Selection for the Fully Non-Fixed Contact Manipulation of Deformable Objects
    L. Han, Y. Zhang, and H. Wang
    IEEE Robotics and Automation Letters (RA-L), 2022
    PDF, demo

    hpp

    Hybrid Adaptive Vision-Force Control Under the Bottleneck Constraint
    L. Han, Y. Zhang, and H. Wang
    IEEE Transactions on Control Systems Technology (T-CST), 2022
    PDF, demo

    hpp

    Adaptive Visual Servoing Shape Control of A Soft Robot Manipulator Using Bézier Curve Features
    F. Xu, Y. Zhang, S. Jing, H. Wang, and W. Chen
    IEEE Transactions on Mechatronics (T-Mech), 2022
    PDF, demo

    Projects

    hpp

    Fruit Collector: Soft Pneumetic Hand Design and Vision-based Grasping
    This skeleton-reinforced soft pneumatic hand designed by topology optimization is capable of picking up a variety of delicate fruits given image input.

    hpp

    Consensus-based Distributed Formation Control and Obstacle Avoidance
    These homogeneous turtlebots can choose between different formations to pass through obstacles.

    hpp

    Cable Assembly: Ends Detection and Vision-based Manipulation
    An LSTM model trained for cable end detection helps non-calibrated adaptive vision based manipulation when self-occlusion happens.

    Academic Services

  • Conference Reviewer: ICRA 2023, 2024, 2025
  • Journal Reviewer: IEEE Transactions on Mechatronics, IEEE Access

    • Teaching

  • TA: 24677 Linear Control Systems, 2023 Fall

    • Coursework

  • 36-700 Probability and Mathematical Statistics: cheatsheet
  • 36-705 Intermediate Statistics: midterm1, midterm2, final cheatsheet
  • 10-725 Convex Optimization: midterm cheatsheet, final cheatsheet