https://www.physicalintelligence.company/blog/pi0

https://arxiv.org/pdf/2304.13705

"[w]e develop a simple yet novel algorithm, Action

Chunking with Transformers (ACT), which learns a generative

model over action sequences. ACT allows the robot to learn 6

difficult tasks in the real world, such as opening a translucent

condiment cup and slotting a battery with 80-90% success,

with only 10 minutes worth of demonstrations. Project website:

tonyzhaozh.github.io/aloha "

https://x.com/RemiCadene/status/1828006677528137977

Does the wearer have to survive?

lmao

We are looking for human level skill on necktie tying, and humans know how to do it without killing the wearer. so yes, the wearer should be alive after this interaction.

I would have thought there were already robots machines that could do this, and have been for a long time?

is this true? can you point to a credible source/demonstration?

Idk, as I said, it's something I would have thought. This should be doable with a relatively simple mechanical contraption. Given that selling pre-tied ties is a thing, one would think that machines to do the tying exist.

That's quite different from the resolution criteria (ie prev. unseen subjects and ties). Closest I have seen is here https://tiebots.github.io/

Well, yes. Using AI to do this seems like making the task more difficult for oneself.