Reward AI showed a humanoid pulling packages off a Cybertruck bed and working on stairs, and says it was trained only on people wearing a sensor glove.
The Reward AI blog says there was no teleoperation and no on-robot experience, and that the model picks up a brand-new task from less than 30 minutes of that human data. The video runs 77 seconds, from boxes coming off the truck bed to the robot on a staircase.
That is what I find more interesting than the delivery itself. Collecting robot data has been the slow and expensive step in this field, because every hour of it needs a robot, an operator and a lab. If a person wearing a glove can replace all three, the data bottleneck gets much smaller.
It’s a demo, and the last few metres from the truck bed to a door are a narrow part of the job. But I think a robot that learns from human hands is the version that scales, and I expect to see the same approach on a lot more tasks soon.