3 ms·

Tesla's Optimus and the Humanoid Robot Race Nobody's Ready For

4 points by Bluestein 1y ago

cranberryturkey 1y ago
I will pay for a robot to assemble Ikea furniture.
- Someone 1y ago
  Reminds me of https://en.wikipedia.org/wiki/Insert_Knob_A_in_Hole_B https://en.wikipedia.org/wiki/Insert_Knob_A_in_Hole_B: “Plot summary Two men on a remote space station receive all of their equipment from Earth unassembled, and must assemble it with only vague and confusing instructions ("composed by an idiot", one says); as a result, it often fails to work properly or at all. They eagerly await the arrival of a sophisticated positronic robot that will repair existing equipment and assemble new ones. Upon its arrival, they discover that the robot has been shipped in 500 pieces with vague, confusing assembly instructions.”
  - cranberryturkey 1y ago
    haha. yeh that's how it would work too. lol
Zigurd 1y ago
Is there even one use case where a humanoid robot has proven to be the best solution? To illustrate the point I would bet money that a laundry folding machine that only folds laundry better, faster, and cheaper will be the best solution for that use case. Lawn mowing? Another robot. Floor cleaning? Done and dusted. If there is a task with enough value to automate, a specialized robot will win that market.
- hollerith 1y ago
  The advantage of the humanoid robot is compatibility with a built environment tailored for humans.
  - Zigurd 1y ago
    Like sitting in the driver's seat of a non-automated car and being a chauffeur? There are one and a half billion non-automated vehicles out there. An obviously gigantic market. And yet I'm pretty sure everyone can see that a humanoid robot driving a car is a ridiculous use case.
- jfengel 1y ago
  The case I have in mind: cleaning hotel rooms. Build me a robot that can clean the toilet, change the sheets, replace the soap, check the mini-fridge, and do it all without breaking people's stuff scattered around the room. This is minimum-wage work which probably costs you $30,000 a year (including taxes, benefits, etc.) A $100,000 robot would replace that cost in just over three years -- one year, if it could work 24 hours a day instead of 8. That seems a huge ask. The tasks are all simple, but currently require AI-complete levels of perception, especially if combined into one machine. I doubt that a specialized robot -- or a series of them -- is likely to be cost-effective. Currently, they don't even use roombas.
einrealist 1y ago
Found a TSLA shareholder.

4 points by Bluestein 1y ago

cranberryturkey 1y ago

I will pay for a robot to assemble Ikea furniture.

Someone 1y ago
Reminds me of https://en.wikipedia.org/wiki/Insert_Knob_A_in_Hole_B https://en.wikipedia.org/wiki/Insert_Knob_A_in_Hole_B: “Plot summary Two men on a remote space station receive all of their equipment from Earth unassembled, and must assemble it with only vague and confusing instructions ("composed by an idiot", one says); as a result, it often fails to work properly or at all. They eagerly await the arrival of a sophisticated positronic robot that will repair existing equipment and assemble new ones. Upon its arrival, they discover that the robot has been shipped in 500 pieces with vague, confusing assembly instructions.”
- cranberryturkey 1y ago
  haha. yeh that's how it would work too. lol

Zigurd 1y ago

Is there even one use case where a humanoid robot has proven to be the best solution? To illustrate the point I would bet money that a laundry folding machine that only folds laundry better, faster, and cheaper will be the best solution for that use case. Lawn mowing? Another robot. Floor cleaning? Done and dusted. If there is a task with enough value to automate, a specialized robot will win that market.

hollerith 1y ago
The advantage of the humanoid robot is compatibility with a built environment tailored for humans.
- Zigurd 1y ago
  Like sitting in the driver's seat of a non-automated car and being a chauffeur? There are one and a half billion non-automated vehicles out there. An obviously gigantic market. And yet I'm pretty sure everyone can see that a humanoid robot driving a car is a ridiculous use case.
jfengel 1y ago
The case I have in mind: cleaning hotel rooms. Build me a robot that can clean the toilet, change the sheets, replace the soap, check the mini-fridge, and do it all without breaking people's stuff scattered around the room. This is minimum-wage work which probably costs you $30,000 a year (including taxes, benefits, etc.) A $100,000 robot would replace that cost in just over three years -- one year, if it could work 24 hours a day instead of 8. That seems a huge ask. The tasks are all simple, but currently require AI-complete levels of perception, especially if combined into one machine. I doubt that a specialized robot -- or a series of them -- is likely to be cost-effective. Currently, they don't even use roombas.

einrealist 1y ago

Found a TSLA shareholder.