Machine Learning-based Control for Underactuated Balance Robots
Feng Han
Department of Mechanical and Aerospace Engineering
Rutgers University
Tuesday, Dec 19 at 9:30am in ITE 336
or via webex
https://uconn-cmr.webex.com/uconn-cmr/j.php?MTID=m1b054263d3e10844bf1c26e486290062
Abstract: Underactuated balance robot is a special class of mechanical systems, which possess more degrees of freedom than the number of control inputs, such as rotary/cart-inverted pendulums, autonomous bicycles, bipedal walkers, etc. Control design for underactuated balance robots needs to take care of trajectory tracking of actuated subsystem and balance of unactuated subsystem, simultaneously. A model-based control design is first reviewed to better understand the system properties and limitations of the existing controls. A Gaussian Process (GP)-based system modeling and control design is then presented. GP is used to obtain the estimation of the system dynamics without the need of prior physical knowledge or successful balance demonstrations. The proposed control decouples the actuated subsystem and unactuated subsystem. Tracking and balance performance are guaranteed in the sense of probability. Additional attractive properties of the design include guaranteed stability and closed-loop performance. Experimental results are used to demonstrate the performance. The proposed control framework can be further extended to a general robotic system with safety requirements and dynamics constraints involved.
Short Bio: Feng Han received the B.S. degree in aerospace engineering from Nanjing University of Aeronautics and Astronautics in 2017, and the M.S. degree in aerospace engineering from Harbin Institute of Technology in 2019, respectively. He is currently pursuing his Ph.D. degree in Mechanical and Aerospace Engineering at Rutgers University and plans to graduate in May 2024. His research interests include dynamics and control, machine learning, and automation science and engineering with applications to robotics and autonomous systems.
Currently, Feng Han is a student member of the American Society of Mechanical Engineers (ASME) and the Institute of Electrical and Electronics Engineers (IEEE). He has received several awards, including the China National Scholarship for Undergraduate, one of Ten Outstanding Young Persons in Nanjing University of Aeronautics and Astronautics, and the First-Class Award of Undergraduate Thesis in Jiangsu Province, China. His paper “Autonomous Bikebot Control for Crossing Obstacles With Assistive Leg Impulsive Actuation,” published in IEEE/ASME Transactions on Mechatronics, received the
For more information, contact: Brandy Ciraldo at brandy.ciraldo@uconn.edu