About Me

I am Zhilun Hu (胡至伦), a Ph.D. student in the School of Mechanical Science and Engineering at Huazhong University of Science and Technology (HUST), where I am fortunate to be advised by Prof. Weiming Shen (IEEE Fellow, Academician of the Canadian Academy of Engineering).

My research focuses on Reliable Robotic Manipulation. I investigate how embodied robotic agents can actively acquire unoccluded visual feedback, verify high-level action proposals against geometric constraints prior to execution, adapt contact-rich policies via visuotactile sensing and reinforcement learning, and recover from anomalous execution states.

我是胡至伦，华中科技大学机械科学与工程学院博士研究生，师从沈卫明教授（IEEE Fellow、加拿大工程院院士）。我的研究聚焦于机器人可靠操作（Reliable Robotic Manipulation），致力于构建涵盖双臂主动视觉伺服感知、柔性物体操作执行前视觉校验、视触觉多模态强化学习自适应以及反射式故障恢复与经验记忆的闭环智能操作体系。

Research Interests

  • Active Perception & Visual Servoing: Dual-arm active viewpoint planning and target–gripper relative visual servoing under occlusion.
  • Vision-Language Robotic Manipulation: Iterative visual-overlay action verification and 3D geometric constraint reasoning for deformable objects.
  • Visuotactile Policy Learning: Cross-modal visuotactile fusion and reinforcement learning policy adaptation for contact-rich manipulation.
  • Failure Recovery & Experience Memory: Online value/risk monitoring, reflexive retraction control, and episodic memory retrieval.

🔥 News

  • [2026.06] Our paper Active Observation for Robust Visual Servoing in Robotic Manipulation has been accepted to IEEE CASE 2026 (Paper ThCT6.3, First Author)!
  • [2026.04] Submitted our dual-arm deformable manipulation work VR-Loop: Iterative Verify-and-Refine Vision-Language Agent for Reliable Dual-Arm Manipulation (Second Author).
  • [2026.01] Initiated two new research thrusts: T-Rex / Sticks (visuotactile policy adaptation on Piper robot) and HELIX (reflexive failure recovery & episodic experience memory).
  • [2025.09] Started my Ph.D. journey in Mechanical Engineering at Huazhong University of Science and Technology (HUST), advised by Prof. Weiming Shen.

📝 Selected Publications & Preprints

All Publications →
IEEE CASE 2026
3D visualization of dual-arm active observation and relative visual servoing under occlusion

Active Observation for Robust Visual Servoing in Robotic Manipulation

Zhilun Hu, Yuqi Cheng, Yuxiang Tan, Wen-pan Li, Weiming Shen

IEEE International Conference on Automation Science and Engineering (CASE), 2026 · Paper ThCT6.3 · First Author

During robotic grasping, the manipulator or gripper frequently obscures the target object just when terminal visual feedback is most critical. This work coordinates an active eye-in-hand camera arm with a grasping arm, combining a two-stage viewpoint strategy, target–gripper relative visual servoing, and occlusion-triggered viewpoint exploration.

Project Page Official Program (ThCT6.3) Publication Page
Under Review
3D visualization of VR-Loop dual-arm deformable rope untangling with verified pulling trajectories

Iterative Verify-and-Refine Vision-Language Agent for Reliable Dual-Arm Manipulation

Yuqi Cheng, Zhilun Hu, Cheng Jiang, Qingyu Peng, Weiming Shen

Manuscript under review · 2026 · Second Author

In deformable rope untangling, an unverified grasp point or pulling vector can irreversibly tighten a knot. VR-Loop introduces a closed-loop Verify-and-Refine Vision-Language Agent that renders proposed dual-arm actions onto image overlays, critiques topological and mask constraints prior to execution, and synthesizes depth-informed 3D trajectories.

Project Page Publication Page

🔬 Research Projects

4 projects
IEEE CASE 2026

Active Observation and Visual Servoing

Active perception · Accepted · IEEE CASE 2026

Coordinating an active RGB-D camera arm with relative visual servoing to preserve target–gripper co-visibility under occlusion.

Technical Focus: Two-stage active viewpoint strategy + target–gripper relative image error feedback + occlusion-triggered viewpoint exploration.

Project Details
Under Review

VR-Loop: Dual-Arm Rope Untangling

Dual-arm manipulation · Manuscript Under Review

Pre-execution visual-overlay verification and geometric refinement for dual-arm deformable rope untangling.

Technical Focus: Closed-loop VLM action critique + image-overlay visual verification + mask & depth geometric projection.

Project Details
Visuotactile RL

T-Rex / Sticks: Visuotactile Policy Adaptation

Visuotactile learning · Ongoing Research

Adapting contact-rich manipulation policies via vision, tactile force fields, and reinforcement learning on a Piper robot platform.

Technical Focus: Visuotactile sensorimotor fusion + reinforcement learning policy adaptation + continuous action-data coupling on Piper hardware.

Project Details
Failure Recovery

HELIX: Failure Recovery and Experience Memory

Failure recovery · Ongoing Simulation Research

Integrating online value assessment, fast reflex feedback, and episodic experience memory to recover from manipulation failures.

Technical Focus: Online value/risk monitoring + reflexive safe retraction + episodic experience memory retrieval in simulation.

Project Details

📖 Educations & Academic Training

  • Present, Ph.D. Candidate in Mechanical Engineering (Robotics & Embodied AI), School of Mechanical Science and Engineering, Huazhong University of Science and Technology (HUST).
    Advisor: Prof. Weiming Shen (IEEE Fellow, Academician of CAE).
    华中科技大学 · 机械科学与工程学院 · 机械工程博士研究生（机器人与具身智能方向）（导师：沈卫明教授）