I am Yuqiang Yang, a master student at South China University of Technology, supervised by Prof. Chenguang Yang. Now I am also working as an intern at Application Innovate Laboratory, Huawei Technologies Co.Ltd.
My research interests are robotic learning from demonstration, mobile manipulation and whole-body control. Specifically, I have been exploring and exploiting the potentiality of the wheeled mobile manipulators in acquiring manipulation skills from human demonstration, such as picking-and-placing or openning the door. Recently I am working on the wholebody control of the wheeled mobile maniputors through SLQ-MPC controller implemented in OCS2. This goal of this project is to enable mobile manipulators to plan and execute the manipulation in real time in a cluttered environment.
South China University of Technology
Supervisor: Prof. Chenguang Yang
South China University of Technology
GPA: 3.94/4.0 (ranked first)
Realtime wholebody control of the mobile manipulator in cluttered environment
Propose and analyse the adaptive MPC theorectically to improve the convergence and optimality in complex environments where multiple non-convex costs or contraints exist
The proposed framework is verified through experiments about picking and placing a cup gracefully in clutted environment.
Learning the coordination motion of mobile manipulators through human demonstration
Advisor: Prof. Chenguang Yang
Propose a whole-body LfD framework through Gaussian Process, which endows the mobile manipulator’s skill learning process with features of large-scale convergence, coordination working and disturbance rejection, after just a few human demonstrations.
An efficient kinesthetic teaching method is devised based on the weighted least-norm (WLN) inverse kinematics solution and an admittance controller, which facilitates human users to guide the mobile manipulator to perform tasks.
The framework allows for human-in-the-loop correction when the whole-body is conducting a task.
Self-balanced smart car with wireless charging capability
Advisor: Dr. An Chen
Design and implement the adaptive wireless-charging algorithm to quickly charge the supercapacitors mounted on the car. A series control sytem from the charing power loop to the charging current loop is designed to stablize the charging power at 30W.
Control the position and orientation of the two-wheeled car to complete a complex race track with different elements such as circles and slopes. The perception infomation mainly comes from the on-board IMU and electromagnetic sensors.
We finished the race in fifth national ranking with a time of 23.8s and won the first prize.
Graceful wholebody pick-and-place for mobile manipulator
Advisor: Dr. Chen Chen
Train GGCNN for the peception of the objects' pose and the grasp quality. The inputs of GGCNN are pointclouds while the ouputs are the grasp quality, width and orientation.
Jointly consider the manipulability, energy, manipulator orientation, path tracking in the QP solvers. The dynamic-weighted QP is implemented based on the tracking error to acheive graceful picking-and-placing.
Programming and Learning Framework:
Python, MATLAB, C/C++, PyTorch, Pybullet
Robotics:
DH, wholebody control, Admittance/Impedance Control, Gravity Compensation, Shared control, teleoperation
Hardware Experience:
Franka, UR10, Mobile Manipulator, Robotiq 2F85, Vicon, Touch X, ATI sensors, STM32