We've got hundreds of thousands of people employed whose job is to pick up a box and then set it down again somewhere else. That's their whole job, they do it all day every day. Maybe we can get a robot that does that. Pick up box. Set box down.
These are still tasks that I don't mind doing myself. I want the clanker to scrub the kitchen floor, paint walls and clean the toilet, perfectly.
But that is unlikely to happen. They'll be used for crowd control (which works because they look intimidating) and surveillance. Perhaps they'll also supervise humans that scrub the kitchen floor. Most likely they'll fail entirely.
If he was wearing colored gloves for hand pose detection, these videos would be suitable for a Vision-Language-Action (VLA) model training set. The robotic companies are making and labeling many such videos.
Probably a good thing he's not doing that - it forces each competitor to come up with an own method of training. Otherwise, they could all just train to hi specific video (kind of like how video card mfg's optimize for the tests when competing against other cards).
Earlier discussion of the original (https://generalrobots.substack.com/p/benjies-humanoid-olympi...) article:
https://news.ycombinator.com/item?id=45609922
We've got hundreds of thousands of people employed whose job is to pick up a box and then set it down again somewhere else. That's their whole job, they do it all day every day. Maybe we can get a robot that does that. Pick up box. Set box down.
I got one for "Fingertip Manipulation"
Platinum medal: tying shoelaces while a three-month old kitten plays with them.
These are still tasks that I don't mind doing myself. I want the clanker to scrub the kitchen floor, paint walls and clean the toilet, perfectly.
But that is unlikely to happen. They'll be used for crowd control (which works because they look intimidating) and surveillance. Perhaps they'll also supervise humans that scrub the kitchen floor. Most likely they'll fail entirely.
If he was wearing colored gloves for hand pose detection, these videos would be suitable for a Vision-Language-Action (VLA) model training set. The robotic companies are making and labeling many such videos.
Probably a good thing he's not doing that - it forces each competitor to come up with an own method of training. Otherwise, they could all just train to hi specific video (kind of like how video card mfg's optimize for the tests when competing against other cards).
[flagged]
AI slop
What if my humanoid does cartwheels, is that too wheel?