Curriculum Vitae

Zhilun Hu (胡至伦)

Ph.D. Student in Mechanical Engineering (Robotics & Embodied AI)
School of Mechanical Science and Engineering
Huazhong University of Science and Technology (HUST), Wuhan, China
Advisor: Prof. Weiming Shen (IEEE Fellow, Academician of CAE)
GitHub: @169884902hzl · Homepage: 169884902hzl.github.io/hzlink-academicpages

Education

• Ph.D. Student in Mechanical Engineering — Huazhong University of Science and Technology (HUST)
  Advisor: Prof. Weiming Shen
  Research Focus: Reliable Robotic Manipulation, Active Perception, Visuotactile Learning, and Embodied AI.

Research Thrusts

My doctoral research investigates closed-loop reliability in physical robot interaction across four interconnected pillars:

1. Active Perception & Visual Servoing: Dual-arm viewpoint optimization and target–gripper relative visual servoing under self-occlusion and environmental obstruction.
2. Pre-Execution Vision-Language Verification: Closed-loop Verify-and-Refine visual grounding and geometric mask projection for dual-arm deformable object manipulation.
3. Visuotactile Policy Adaptation: Cross-modal fusion of vision and tactile force fields with reinforcement learning for contact-rich insertion tasks.
4. Failure Recovery & Episodic Memory: Online value/risk assessment, reflexive retraction feedback, and experience memory retrieval after execution anomalies.

Publications & Preprints

1. 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. (Accepted · Paper ThCT6.3 · First Author)
   [Official Program] [Project Page]

2. 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)
   [Project Page]

Research Projects

• Active Observation and Visual Servoing — Dual-arm active RGB-D viewpoint planning and relative visual servoing under occlusion (IEEE CASE 2026, Accepted).
• VR-Loop: Dual-Arm Rope Untangling — Iterative visual-overlay verification and depth-informed geometric constraints for deformable manipulation (Under Review).
• T-Rex / Sticks: Visuotactile Policy Adaptation — Multimodal visuotactile policy adaptation and reinforcement learning on the Piper robot platform (Ongoing Research).
• HELIX: Failure Recovery and Experience Memory — Online value assessment, reflex feedback, and episodic experience memory for recovery from unsuccessful actions (Ongoing Simulation Research).