Researchers propose framework that trains robots to mimic human actions in captioned videos

In a newly published preprint paper, researchers propose a framework that translates videos of human demonstrations to commands robots can understand.