Abstract: In this talk, we cover four aspects of research on general manipulation for assistive robotics : i), efficiently adapting large-scale models to downstream scene affordance understanding tasks, especially in daily living scenarios where gathering multi-task data involving humans requires strenuous effort; ii), effectively learning robot action trajectories with our flow matching policy; iii) sim-to-real physics learning for deformable manipulation and robot-assisted dressing; iv) multi-modal (vision, tactile, audio) learning.
Bio: Fan is a senior research scientist and a project leader at Honda Research Institute EU. He is also a visiting researcher at Imperial College London. Previously Fan was an Eric and Wendy Schmidt AI in Science Postdoctoral Fellow funded by Schmidt Futures, and he received his Ph.D. degree from Imperial College London (advised by Prof. Yiannis Demiris, Personal Robotics Lab). Fan's research interests cover many aspects of machine learning algorithms applied to general robot manipulation. His prior research has been accepted at Science Robotics and T-RO, and he has been awarded The UK Best PhD in Robotics Award 2020 1st place held by Queen Mary, and Best Research Paper (Early Career Researcher), 2025 AI & Robotics Research Awards held by TAS Hub and Responsible Ai UK.