3 articles · Last covered
Reward AI’s latest OM-1 video features delivery, stairway picking and household tasks, following its September debut of a policy trained on human demonstrations.
From a handheld GoPro gripper to Sunday’s Skill Capture Glove, Reward AI’s Omnibody Hand and X Square’s TwinDEX, a growing school of robot development starts with the device humans use to teach the machine.
Spun out of Stanford’s DexCap project, Reward AI has introduced OM-1, an embodied foundation model trained exclusively on wearable human data that transfers zero-shot across arms and humanoids.
