RialTo created robust policies in a real-world kitchen & labs settings on tasks like placing a mug on a shelf & opening shelves. RialTo provides a 67% improvement over imitation learning approaches w/the same number of demonstrations.

→ View original post on X — @mit_csail,