Dr Yanlong Huang

Position: University Academic Fellow
Areas of expertise: Imitation learning; optimal control; reinforcement learning; motion planning; robotics.
Email: Y.L.Huang@leeds.ac.uk
Location: 6.22 E C Stoner Building

Profile

I am an university academic fellow (equivalent to assistant professor) at the school of computing, University of Leeds. My research mainly focuses on imitation learning, optimal control, reinforcement learning, motion planning and their applications to robotic systems. I received my Ph.D. degree from Institute of Automation, Chinese Academy of Sciences, Beijing, China. After that, I carried out my research as a postdoctoral researcher in Max-Planck Institute for Intelligent Systems and Italian Institute of Technology.

Updates

Y. Huang and D. G. Caldwell. A linearly constrained nonparametric framework for imitation learning, IEEE International Conference on Robotics and Automation (ICRA), Pairs, France, 2020, Accepted. Paper Video

Research interests

I have been focusing on developing generic approaches for robot skill learning and optimization. Specifically, my research topics comprise imitation learning, optimal control, reinforcement learning and motion planning. Please see part of my work as follows.

Imitation learning, as a promising branch in robot learning, is aimed at extracting relevant motion patterns from human demonstrations and subsequently applying these patterns to new situations. Recently, we have developed a novel framework which can solve various key issues arising from imitation learning. Paper 1 Paper 2 Paper 3

While learning Cartesian positions or joint angles suffices for many applications, the end-effector orientation is required in many others. We have provided a novel solution for orientation learning and adaptation in Cartesian space, which allows for learning multiple orientation trajectories and adapting learned orientation skills to new situations (e.g., via-point and end-point), where both orientation and angular velocity are addressed. Paper 1 Paper 2

In order to solve the problem of learning and adapting human skills in both Cartesian and joint spaces simultaneously, we developed a hybrid adaptive imitation learning framework, providing a more flexible solution compared with traditional approaches that are built in either Cartesian space or joint space. Paper 1 Paper 2

When I worked in Max Planck Institute for Intelligent Systems, we proposed a pure data-driven learning framework in table tennis robot, where imitation learning offered initial solutions for robot striking movements while reinforcement learning provided online adjustments. Before that, in the Institute of Automation, Chinese Academy of Sciences, we developed a combined learning framework comprising supervised learning and active feedback learning, which was employed to a five degrees of freedom table tennis robot to address the problem of returning arbitrary incoming balls to a desired position. Paper 1 Paper 2

<h4>Postgraduate research opportunities</h4> <p>We welcome enquiries from motivated and qualified applicants from all around the world who are interested in PhD study. Our <a href="https://phd.leeds.ac.uk">research opportunities</a> allow you to search for projects and scholarships.</p>

Projects

<li><a href="//phd.leeds.ac.uk/project/665-constrained-robot-skill-learning-from-few-examples">Constrained Robot Skill Learning From Few Examples</a></li>

Dr Yanlong Huang

Profile

Updates

Research interests

Projects

Publications:

My five most recent selected publications

My other selected publications

Journal articles

Conference papers

Presentations (conference/workshop etc)

Conference presentations

Books

Chapters

Reports

Internet publications

Performances

Compositions

Exhibitions

Posters

Artefacts

Designs

Patents

Scholarly editions

Software / codes

Thesis / dissertations

Others

Datasets

Figures

Filesets

Medias

Working papers