Hao Wu | 吴浩 I am a bachelor's graduate majoring in Mechanical Engineering from Zhejiang University. I received B.Eng with honors from Excellence Talent Training Program in 2024. Now, I work as a research assistant in the Robotic Perception and Grasp Lab. I am honored to be advised by Prof. Huixu Dong, Prof. K. Jimmy Hsia, and Dr. Changhong LINGHU. And I am privileged to work closely with Prof. Tiefeng Li, Dr. Yanzhe Wang, and Mr. Haotian Guo. Currently, I am actively seeking opportunities for graduate studies. Please feel free to contact me if you require any additional information or have further inquiries. Email  /  Google Scholar  /  Github /  WeChat /  CV /  Portfolio

My research interests are broadly on Robotic Locomotion, Perception and Manipulation, with particular interests in the development of biomimetic robots, soft manipulators, and multimodal haptic sensors. My work involves the integration of advanced materials, mechanical engineering, and sensor technology to create adaptive robotic systems capable of sophisticated interactions with their environment. I am passionate about exploring materials and mechanisms that endow robots with the dexterity to manipulate objects with the same finesse as humans. Representative works are highlighted. Click here for my portfolio.

	Enabling Tunable Stiffness, Adhesive Grasping, and Interaction-driven Reconfiguration: A SMP-Enhanced Fin-ray Gripper Haotian Guo*, Hao Wu*, Yanzhe Wang*, Yaoting Xue, Tuck-Whye Wong, Tiefeng Li, and Huixu Dong† (*equal contribution) Paper / Video Soft Robotics Accpeted Take advantage of the thermoresponsive stiffness and reversible adhesion of Shape Memory Polymers (SMPs) to augment the grasping modality in fin-ray gripper design. Validate the high load capacity, good shape adaptability and adhesive grasping capability with gripper successfully lifting objects of arbitrary shapes and weights, grasping large-diameter sphere, concave, planar, and tiny items. Actively modify the gripper’s conformation with intended engagement with objects or environment and implement human-inspired strategies to manipulate tools or handle inaccessible objects.
	Tactile-integrated FlexiRay: Breaking Planar Limits by Harnessing Large Deformations for Flexible, Full-Coverage, Human-like Multimodal Sensing Yanzhe Wang*, Haotian Guo*, Hao Wu*, and Huixu Dong† (*equal contribution) Paper / Video / Arxiv Nature Communications In Revision Introduce the vision-based tactile sensor to fin-ray gripper which facilitates multimodal perception of contact force, texture, slippage, temperature, and proprioception without compromising the its adaptability and flexibility. Optimize the positioning and orientation of mirrors with CMA-ES algorithms, ensuring that the internal camera can reliably acquire full-coverage imagery through mirror reflection even during the instances of significant deformation. Illustrate the performance of the Tactile-integrated FlexiRay in recognizing multimodal contact information in specific grasping and operational tasks through learning-based approach.
	Integrating Biomimetic Synergy with Linkage-Driven Mechanisms: An Anthropomorphic Hand for Versatile Grasping and Manipulation Hao Wu, Zhaohui Lin, Yanzhe Wang, Haotian Guo, and Huixu Dong† Paper / Video / Poster / Graduation Thesis / Patent Conduct biomechanical analysis of human hand synergistic movements to provide reasonable simplifications of its intricate and compact nature. Develop an integrated linkage-driven anthropomorphic hand featuring 19 joints and 11 active actuators. Novel linkage mechanisms are proposed with optimized geometric parameters to achieve dexterous movements and resemble natural human gestures. Perform grasping and manipulation experiments to demonstrate the hand’s capability to adapt to various objects and manipulate tools for specific tasks, validating its human-like operational strategies.
	SA-DEM: Dexterous Extrinsic Robotic Manipulation of Non-Graspable Objects via Stiffness-Aware Dual-Stage Reinforcement Learning Yanzhe Wang, Wei Yu, Hao Wu, Haotian Guo, Huixu Dong† Paper IEEE Transactions on Automation Science and Engineering In Revision A novel dual-stage reinforcement learning method that integrates object stiffness for the first time to achieve flexible, autonomous planning for non-grasping manipulation. A unified 3D spatial observation representation and a consistent actor-critic framework that effectively addresses pose decision-making and action interaction in extrinsic dexterous manipulation. A data-driven learning strategy for the high-level agent that integrates discrete sparse action decisions into a continuous reward representation, significantly enhancing training performance. The trained policies demonstrate exceptional perceptual decision-making capabilities and dynamic manipulation skills, enabling it to autonomously determine target poses for objects of varying stiffness and execute diverse actions such as pushing, poking, and flipping.
	Design and Application of Gecko Bio-inspired Climbing Robot Based On SMP Adhesives Supervisor: Prof. K.Jimmy Hsia; Collaborator: Changhong LINGHU, Yan Li etc. Science Robotics In Preparation Integrate array embedded heaters with E44 epoxy Shape Memory Polymer (SMP) as smart adhesive devices to achieve precise heat transfer and rapid thermo-responsive actuation. Validate the potential to use the SMP R2G adhesive fibrils as a soft gripper to grip heavy and rough objects, which will further serve as the footpads of a quadrupedal robot for climbing and withstanding heavy payloads on various surfaces.
	Developments of Multi-Modal Vision-Based Tactile Perception for Robotics Dexterous Manipulation Supervisor: Prof. Huixu Dong; Collaborator: Haotian Guo In Preparation Design a vision-based tactile sensor with omnidirectional 3D perception capability, which enables multimodal perception of contact pressure, texture, slippage, and vibrations. The sensor is composed of a silicone elastomer coated with a reflective surface, PDMS supporter, optical waveguides for internal illumination and response for high-frequency vibrations, camera, LEDs and phototransistors. Implement unsupervised learning to compress the tactile features and plan to further apply tactile model predictive control for dexterous manipulation tasks.
	Development of a Tendon-Driven Dexterous Finger with Variable Length Capability for Advanced Manipulation and Rehabilitation Applications Supervisor: Prof. Huixu Dong; Collaborator: Haotian Guo, Yanwen Zou, Tianyang Qian In Preparation Design a tendon-driven dexterous finger with rolling joints capable of bidirectional movements and length adjustment. Establish kinematic and static models to characterize the finger’s motion, facilitating precise control and actuation. Conduct experiments to validate the feasibility of the finger as a component in the dexterous hand, rehabilitation exoskeleton, and sensor glove.
	Individual Projects for Interest Acknowledegement to: Chengkun Zhang, Shijie Zhou, Yujie Kong, Koala Workshop, Junyao Zhuge etc. Construct the hardware and software platform for an Underwater Vehicle. My work involves structural design, modeling and simulation, manufacture and assembly, sensor integration, remote control, and underwater validation Accomplish the mechanical fabrication of a 13-DOF Bipedal Robot and achieve motion control with simple gaits. Replicate the appearance and locomotion of spiders and develop a Hexapod Robot that facilitates diverse movements on various road surfaces through the integration of 18 servomotors and 3D-printed components. Develop a Rubik’s Cube Robot capable of automatically recognizing individual color blocks and solving for the solutions, consequently using six stepper motors to restore the cube within seconds. Member of RoboMaster University Championship. Design the hardware and software of an electromagnetic racing car.

I have gained comprehensive theoretical knowledge as well as practical capabilities during my undergraduate studies. Below is a list of projects that I have led.

	Course: Comprehensive experiment of Mechatronic System Conduct visual measurement of gear profiles with the software Vision Pro. Utilize a camera and a single-axis rotary stage to generate point cloud data, thus facilitating the three-dimensional reconstruction of workpieces. Achieve high-precision eye-to-hand calibration of the robotic system.
	Course: Robotics Calculate the forward and inverse kinematics of a six-axis robotic arm with the Denavit-Hartenberg method. Implement trajectory planning in both joint space and Cartesian space. Design controllers for the dynamic control of the robotic arm with Simulink, including gravity compensation PD controller, inverse kinematics controller, and force-position hybrid controller.
	Course: Wheeled Robot Technology and Reinforcement Practice Implement A* and Rapidly-exploring Random Tree *(RRT*) algorithms for global path planning in a known map. Design a Dynamic Window Approach(DMA) algorithm for online obstacle avoidance. Obstacles can be added to the map after path planning. The video demonstrates the performance of A* and RRT* path planning & DWA in evading both static and dynamic obstacles.
	Course: Analysis of Flow Field Simulate the process of an object falling into the water and forming a Worthington jet in ANSYS. Investigate the interfacial flow dynamics and the impact on Worthington jet heights following the impact of objects on liquid surfaces under varying conditions of mass, velocity, size, and fluid medium.

	Zhejiang University (ZJU) 2020.9 - 2024.6 Bachelor of Mechanical Engineering, Excellent Talent Training Program Overall GPA : 3.98/4 (89.93/100) Advisor: Prof. Yicong Gao, Prof. JiangXin Yang
	Nanyang Technological University 2023.07 - 2023.09 Summer Research Intern Research Advisor: Prof. K Jimmy Hsia, Prof. Huajian Gao, Dr. Changhong LINGHU
	Robotic Perception and Grasp Lab, Zhejiang University 2024.08 - Present Research Assistant Research Advisor: Prof. Huixu Dong, Prof. Tiefeng Li
	University of Hong Kong 2021.08 Visiting Student

Last updated: December.1, 2024
© 2024 Hao Wu