Speaker: Tao Chen MIT CSAIL
Time: 2023-04-20 10:00-2023-04-20 11:00
Venue: C19-1 or Online: https://us06web.zoom.us/j/85822472330?pwd=VU1DTE1wek0zWHlOMlpaQko1dXNnZz09
In recent years, there has been significant progress in robotics grasping, with many techniques being commercialized and utilized in factories and warehouses. However, comparatively, dexterous manipulation research has received much less attention due to its complexities in control, sensing, hardware design, etc. In this talk, I will share our initial effort in developing a learning pipeline that efficiently learns generalizable contact-rich skills from raw sensory data using a low-cost hardware system costing less than $5000. Our system has achieved zero-shot sim-to-real transfer on dynamic and contact-rich tasks (in-hand object reorientation), and is capable of reorienting hundreds of geometrically distinct objects.
Tao Chen is a Ph.D. student at MIT CSAIL, advised by Prof. Pulkit Agrawal. His research interests include all aspects of robot learning, with a recent focus on dexterous manipulation. His work on in-hand object reorientation won the best paper award at CoRL 2021.