We've had isolated demos of this and that for many decades, all of varying difficulty of course, but I for one definitely agree with the post. I don't feel like I could go and buy a generic home assistant robot that can perform a variety of basic tasks anytime soon. Demos are nice, but they're kinda like news about new batteries, always a few years away from the real thing.

I'd bet many more than a few years away from a robot that can arbitrarily navigate your house, gather laundry from several baskets, the floor, etc. Divide into whites/colors or warm/cold loads if desired. Load them into an arbitrary washer, measure out detergent, use the right washer settings for the given load. Then come back and move them to the dryer, then come back and unload, fold, put away, etc. And don't get me started on sheets.

It's much easier than it should be to fool people with demos.

All whilst safely navigating around people, pets, babies, spills etc.

It's one thing for a little robot vacuum to do. Whole other when it's a very large, very heavy humanoid robot.

Absolutely. And that demo was so bad, the robot was actually shitty at folding clothes (no offense to the robot personally, of course...)

solid state batteries starting to actually hit the market, as in you can buy one today.

https://yoshinopower.com/collections/portable-power-stations

10 years ? 20 ?

https://mobile-aloha.github.io/

(note: in some videos the robot is teleoperated)

Here is the earlier demo [1965]

https://www.youtube.com/watch?v=z4bdaN6MRT8

Haven’t most of those been teleoperated?

https://www.physicalintelligence.company/blog/pi0 could all be lies and I've been duped, but I'd like to believe not

but how far are we from some bot ordering you a breakfast burrito on DoorEats and having it delivered? I can 100% envision a bug in one of the current crop of "AI" products resulting in mass burritoing.

The bot already does that, you command it through your web or mobile interface

turn this into something commercial is the real trick