MIT Technology Review @techreview
A virtual robot arm has learned to solve a wide range of different puzzles—stacking blocks, setting the table, arranging chess pieces—without having to be retrained for each task. https://t.co/iZVvPJwVC3 — PolitiTweet.org