4 ms·

To do what? Even the humanoid bots don't have clear use cases imo

tialaramex 2y ago
Boston Dynamics has video of the humanoids tasked to unpack or pack freight. You could hire people to come in and pack or unpack trucks, or, you could have a humanoid robot. Truck pulls up to the loading dock, opens doors, robot takes all the boxes out and puts them on the conveyor, or takes boxes off a conveyor and puts them in the truck. Doesn't matter when the truck arrives, the robot isn't at lunch break or on holiday or gone home for the day, so like if you'd hired 3-4 shifts you can always do this. But while 3-4 shifts makes sense if you really do this 24/7, if you do it once a month but with no predictable schedule you cannot hire for that. Likewise if every truck has 85cm cube boxes in it, you don't need humanoids you can purchase a solution. But the Boston Dynamics robots, just like human workers don't care that the boxes are a bit different, within reason. The robots are less adaptable than humans, but they're much more adaptable than previous non-human approaches.
- vonneumannstan 2y ago
  I think my larger point is that it seems doubtful that most use cases need a humanoid form. Like is there any reason to think a specialized unloading bot is worse than a humanoid bot in this case?
  - tialaramex 2y ago
    There are loads of edge cases where of course your specialist bot can be geared for that edge case, but each such bot is different. So Boston Dynamics sells fifty companies their $1M robots (example, I have no idea of actual prices) meanwhile a custom robot costs $1M because of the extensive design and engineering work for each such custom robot. If you were buying a lot of them they'd be cheaper, but you aren't and they know this custom design won't suit the next customer. If you were conjuring into existence arbitrary environments, Boston's humanoids lose, but the reason they picked humanoids† is that in these environments the existing workforce are humans. Humans don't mind the weird ledge needed because of a legal change in 1896, or the uneven weight distribution in the boxes, they adapt and so can Boston's robots. Humanoids couldn't fit down a 6cm diameter pipe, a custom robot could, but, since the people doing the job last month were humans it's not in a 6cm diameter pipe, the humans doing it required an access hatch, and the humanoid robot can operate that hatch too. † You can make a similar argument for the canine robots. People already use dogs so more of our world is tweaked to make it suitable for dogs.
    - vonneumannstan 2y ago
      Seems at most a transitory situation until places retool for dedicated bot environments. Its simply less efficient to cater to humanoid workers.

Boston Dynamics has video of the humanoids tasked to unpack or pack freight. You could hire people to come in and pack or unpack trucks, or, you could have a humanoid robot. Truck pulls up to the loading dock, opens doors, robot takes all the boxes out and puts them on the conveyor, or takes boxes off a conveyor and puts them in the truck. Doesn't matter when the truck arrives, the robot isn't at lunch break or on holiday or gone home for the day, so like if you'd hired 3-4 shifts you can always do this. But while 3-4 shifts makes sense if you really do this 24/7, if you do it once a month but with no predictable schedule you cannot hire for that. Likewise if every truck has 85cm cube boxes in it, you don't need humanoids you can purchase a solution. But the Boston Dynamics robots, just like human workers don't care that the boxes are a bit different, within reason. The robots are less adaptable than humans, but they're much more adaptable than previous non-human approaches.

vonneumannstan 2y ago
I think my larger point is that it seems doubtful that most use cases need a humanoid form. Like is there any reason to think a specialized unloading bot is worse than a humanoid bot in this case?
- tialaramex 2y ago
  There are loads of edge cases where of course your specialist bot can be geared for that edge case, but each such bot is different. So Boston Dynamics sells fifty companies their $1M robots (example, I have no idea of actual prices) meanwhile a custom robot costs $1M because of the extensive design and engineering work for each such custom robot. If you were buying a lot of them they'd be cheaper, but you aren't and they know this custom design won't suit the next customer. If you were conjuring into existence arbitrary environments, Boston's humanoids lose, but the reason they picked humanoids† is that in these environments the existing workforce are humans. Humans don't mind the weird ledge needed because of a legal change in 1896, or the uneven weight distribution in the boxes, they adapt and so can Boston's robots. Humanoids couldn't fit down a 6cm diameter pipe, a custom robot could, but, since the people doing the job last month were humans it's not in a 6cm diameter pipe, the humans doing it required an access hatch, and the humanoid robot can operate that hatch too. † You can make a similar argument for the canine robots. People already use dogs so more of our world is tweaked to make it suitable for dogs.
  - vonneumannstan 2y ago
    Seems at most a transitory situation until places retool for dedicated bot environments. Its simply less efficient to cater to humanoid workers.