We all want robots to clean our dishes, wash our clothes, and do other household chores. Getting robots to master dexterous hands with real-time hand-eye coordination is challenging. This video from the Robot Locomotion Group shows a self-supervised correspondence in visuomotor policy learning.
Thanks to this approach, robots can do a wide variety of tasks with minimal human effort. For some of them, it takes just a few minutes to learn a new task reliably.
[HT]