A robot learned dexterous hands from minutes of play
Ruoshi Liu says minutes of human play, with no task demonstrations and no sim2real, were enough to train a dexterous manipulation policy. Jitendra Malik picked it up as his answer to what world models should actually be trained on.
Shown this week
Robot learns dexterous manipulation from minutes of human play
Ruoshi Liu released Dexterity from Supervised Play, which he says turns minutes of human play into both a world model and a dexterous manipulation policy. No task demonstrations and no sim2real (train in simulation, then run on the real robot).
World Models" are the rage for very good reasons, but then what is the right sensorimotor experience for training these models?
Out in the world
Origin Autonomy says its drywall robot works at 3x human speed
Origin Autonomy left stealth with $20M and OG-1, a robot that it says finishes drywall autonomously at three times human speed, per Lukas Ziegler. The post says it applies compound, sands, primes and handles corners, pillars and soffits up to 14 feet using force control, with first robots on New York job sites within 10 months of founding.