ZHUO LI
ZHUO LI
I’m a first-year Ph.D. student in Robotics at The Chinese University of Hong Kong (CUHK). I received a master’s degree from Huazhong University of Science and Technology (HUST). My research focuses on the intersection of Robotics and AI, specifically on Humanoid Robot Grasping and Manipulation, as well as Multimodal Large Models and Embodied Intelligence. I was previously an intern at UBTech Robotics, where I contributed to Walker humanoid robot projects.
Education
The Chinese University of Hong Kong 2023.09 – Now
P.h.D Robotics
Supervisor: Prof. Fei CHEN
Huazhong University of Science and Technology 2020.09 – 2023.07
Master Degree, Robotic Engineering
GPA: 3.63/4.0 (Core Courses: Robotics (95), Robot Operating System (90))
Supervisor: Prof. Shiqi LI
Master Thesis: Grasp Synthesis and Motion Planning for Dexterous Manipulation of Humanoid Robots
Research Interests
My current research covers the following topics:
- Dexterous Roboitc Grasping
- Humanoid Bimanual Manipulation
- Multimodal Large Models for Humanoid Reasoning
Publications
Journal Papers:
1.
Language-Guided Dexterous Functional Grasping via Large Language Models for Humanoid Manipulation Zhuo Li, Junjia Liu, Fei Chen Submitted to IEEE Transactions on Automation Science and Engineering 2024.
1.
Planning Multi-fingered Grasps with Reachability Awareness in Unrestricted Workspace
Zhuo Li, Shiqi Li, Ke Han, Xiao Li
Published in Journal of Intelligent & Robotic Systems 2022 (SCI, IF=3.129).
2.
TCDG: Target-driven Collision-aware Dexterous Grasping for Novel Objects in Clutter
Zhuo Li, Shiqi Li, Ke Han, Xiao Li
Submitted to Intelligent Service Robotics 2022 (SCI).
Conference Papers:
Patents:
1.
A Flexible Fixture for Automatic Assembly of Car Seat Slides
Gu Wenbin, Zhuo Li, Wang Yi, Zhang Zeliang
Chinese invention patents, 2020. [Link]
2.
An Energy Consumption Measurement Device for Machining Equipment
Gu, Wenbin, Zhuo Li, Yuan Minghai, Yuxin Li
Chinese invention patents, 2019. [Link]
3.
An Automatic Assembly Device for Car Seat Slide
Gu Wenbin, Zhuo Li, Wang Yi, Li Yuxin
Chinese invention patents, 2019. [Link]
4.
An Energy-saving Modeling Method for Energy Consumption of CNC Machine Tools in Machining Process
Gu, Wenbin, Zhuo Li, Yuxin Li
Chinese invention patents, 2019. [Link]
Research Project
1. General Object Grasping for Humanoid Multi-fingered Hands in Unstructured Environments
Team leader, HUST & UBTECH Intelligent Humanoid Service Robots Joint Lab, 2020.12-present
● (Accepted by JINT) Planning Multi-fingered Grasps with Reachability Awareness in Unrestricted Workspace
Motivated by the stringent requirements of unstructured real-world where a plethora of unknown objects reside in arbitrary locations of the surface, we propose a reachability-aware multi-fingered grasp planning framework that can synthesize feasible grasp configurations for novel objects in an unrestricted workspace. Unlike most works that predict if a proposed grasp configuration within the restricted workspace will be successful solely based on grasping stability, our approach further learns a grasping reachability evaluator that estimates if the grasp configuration is reachable or not from robot’s own experience. This eliminates the need for iterating over grasp candidates with motion planning algorithms at run-time to find a reachable grasp and greatly improves multi-fingered grasping efficiency in an unrestricted workspace
- Video:
● Closing the Loop for Multi-fingered Grasping: A Real-time Generative High-DoF Grasp Synthesis Approach
- Enabling autonomous robots to interact in unstructured environments with dynamic objects requires manipulation capabilities that are closed-loop and reactive. However, most existing multi-fingered grasping approaches only perform one-shot grasp detection and thus cannot learn dynamic correcting behaviors that respond to changes in the environment. In this work, we propose a real-time, generative high-DoF grasp synthesis method which can enable closed-loop reactive multi-fingered grasping.
- Video:
● TCDG: Target-driven Collision-aware Dexterous Grasping for Novel Objects in Clutter
- Recent advances in multi-fingered robotic grasping have enabled fast 6-Degrees-Of-Freedom (DOF) single object grasping. Multi-fingered grasping in cluttered scenes, on the other hand, remains mostly unexplored due to the added difficulty of reasoning over obstacles which greatly increases the computational time to generate high-quality collision-free grasps. In this work, we address such limitations by proposing a novel target-driven collision-aware dexterous grasping approach that achieves single-shot recognition for novel objects and requires only single planning for robust collision-free multi-fingered grasping in cluttered environments.
- Video:
2. Online Trajectory Re-Planning for Redundant Anthropomorphic Arms
Participant, HUST & UBTECH Intelligent Humanoid Service Robots Joint Lab, 2020.05-2020.11
● Hybrid Trajectory Replanning-based Dynamic Obstacle Avoidance for Physical Human-Robot Interaction
Interactive robots are required to have rapid reaction generation capabilities for preventing undesired contact and collisions with dynamic obstacles, such as human arms or moving objects. In this work, we proposed a hierarchical replanning framework that rapidly modulates the ongoing trajectory to help 7-DoF redundant manipulators avoid dynamic obstacles in physical human-robot interaction.
- Video:
3. Humanoid Dexterous Manipulation
Participant, HUST & UBTECH Intelligent Humanoid Service Robots Joint Lab, 2019.10-2020.03
● Go Fetch: Mobile Dexterous Manipulation in Unstructured Environments
- With humankind facing new and increasingly large-scale challenges in the elderly care and domestic spheres, automation of the service sector carries a tremendous potential for improved efficiency, quality, and safety of operations. Mobile service robotics can offer solutions with a high degree of mobility and dexterity, however these complex systems require a multitude of heterogeneous components to be carefully integrated into one consistent framework. This work presents a mobile dexterous manipulation system that combines the state-of-the-art perception, localization, navigation, motion planning and dexterous grasping skills into one common workflow for autonomous objects delivery applications in unstructured indoor environments.
- Video:
● Wheel-hand Coordinated Motion Planning for Mobile Manipulation
In this work, we developed a wheel-hand coordinated motion planning algorithm for wheeled mobile robots that can simultaneously plan the path of the mobile base and the path of the end-effector while performing mobile manipulation tasks.
- Video:
Working Experience
● Assistant Lecturer, Huazhong University of Science and Technology, 2021
- Instructor of ME101 Human-machine Collaboration and Interaction (Spring).
- Taught Robot Operating System (ROS) fundamentals, such as Gazebo, Rviz and MoveIt
● Research Intern, UBTECH Robotics Co., Ltd, 2021
- I conducted a three-month research internship at UBTECH Robotics in 2021. During the internship, i mainly engaged with Humanoid Robot Innovation Center (HRIC) on realizing semantic general object grasping for humanoid service robots in domestic scenarios. We developed a semantic multi-fingered grasping framework that comprises a YOLO-based semantic segmentation module and a CNN-based planar grasp point detection module. The deveolped framework eventually achieved a 90.1% semantic grasping success rate on 40 unknown household objects.
- Video:
Awards
- First-class Academic Scholarship, HUST, 2021
- Best Oral Presentation, IEEE 6th International Conference on Control, Robotics and Cybernetics, 2021
- Global Runner-up for Hackathon Humanoid Robot Challenge, World Artificial Intelligence Conference (WAIC), 2020
- Outstanding Graduate, HHU, 2020
- China National Scholarship, Ministry of Education in China, 2018
- Science & Technology Scholarship, HHU, 2018, 2019, 2020
- Academic Excellence Scholarship, HHU, 2018, 2019, 2020
- Pacemaker to Merit Student, HHU, 2018
- Outstanding Individual in Social Activities, HHU, 2017