6 ms·

That is a good point; Musk does sometimes announce things that essentially serve as a placeholder for "we're going to do this thing once the technology improves

That is a good point; Musk does sometimes announce things that essentially serve as a placeholder for "we're going to do this thing once the technology improves enough to make it feasible", and this is likely more of the same. But "useful humanoid robot" is way more forward-looking than anything else he's done this with; the closest thing would be the whole "full self-driving without LIDAR" thing, which keeps getting pushed back and is likely not possible with current technology. And building a humanoid robot that can actually do stuff is probably an order of magnitude more difficult than that. If I had to make a bet on when it was possible, I'd say 25 years. Maybe 10 if Tesla goes all-in on robots and spends tens of billions on it, Manhattan project style.

OnlineGladiator 5y ago
> If I had to make a bet on when it was possible, I'd say 25 years. Maybe 10 if Tesla goes all-in on robots and spends tens of billions on it, Manhattan project style. I have a decade of experience in robotics (including both humanoid robots and self-driving cars) and I think this is optimistic. Robotics is a wonderful field if you like to work on hard research problems. It is a terrible field if you're trying to invent a profitable product.
- pontusrehula 5y ago
  I think industrial robotics is a well established and profitable field.
  - weego 5y ago
    Industrial robotics focuses on single task, highly specialised machines that have few constraints on form, size etc. This is not that. What tesla are conceptually pitching here (let's ignore that they'll deliver something that's pretty useless by comparison to what they're hoping you're assuming they mean) is a not-in-our-lifetimes leap in robotics in multiple ways
  - squarefoot 5y ago
    True, but humanoid robots have little to do with industrial applications. I'm not bashing Tesla or Musk in any way, but this seems more of a mix of PR and research. Robotics is indeed the future, there will be huge demand in a few years, and of course they want to keep a foot in that door, so they need a project which today allows them to say "we're doing robotics too", tomorrow could evolve in big "ooohhh!" and in the meantime let them build experience in the field. In due time all money spent in this field is going to return. In my opinion this robot will probably evolve either into a companion aid for lonely/ill people or maybe a future Tesla luxury car chaffeur, rather than doing any serious mechanical work. For most uses, today a wheeled/tracked platform with grabbers is still cheaper, stronger, better. The knowledge accumulated during the journey to get there however will represent the most valuable asset.
  - fragmede 5y ago
    that Tesla is in the process of building up expertise in. The Fremont, Nevada, China, and German Tesla factories are all filled to the brim with industrial robots.
  - _trampeltier 5y ago
    That 100% something else. Industrial robots are almost allways in a cage for safety. I know there are 'cobots' now, but I just never saw one, except in showcases. Industrial robots just do a fixed task and are dumber than a slice bread. Industrial robots are super heavy for what they can lift. Even they are very efficient with energy, they are still allways on main grid / no batteries. The huge task is anyway not to build a robot. The huge part is to make the robot (or any device) smarter than a slice bread.
- cush 5y ago
  If solid state battery tech takes off, it's entirely possible within a decade.
  - HWR_14 5y ago
    > If solid state battery tech takes off, it's entirely possible within a decade. Your contention is that its power consumption that makes a general purpose humanoid robot impossible?
    - cush 5y ago
      Yes. Great conversation.
- tessierashpool 5y ago
  Musk isn’t actually looking for profitable products. He’s trying to be the king of vaporware. Remember the Cybertruck? Or how he said he could control pigs by putting microchips in their brains? Or how Tesla was all in on Bitcoin until it wasn’t? He probably doesn’t even do it to bump the stock price. It’s probably just about doing whatever will impress Grimes for a little while. Then they both smoke so much weed they forget the previous never-gonna-happen idea and come up with a brand new one.
  - tessierashpool 5y ago
    people can downvote this all they want, but if you’ve ever had a friend who smoked too much weed, or fallen into the habit yourself, you’ve seen the same behavior that Musk exhibits. too many ideas, not enough follow-through, and the ideas tend to be wild and unrealistic —- he’s a textbook case.
    - fitzie 5y ago
      Elon is willing to commit a lot of capital and effort into his ideas. the production capabilities of Tesla are improving at a significant clip, and incorporating many state of the art techniques. Elon's approach to innovation is a lot of hype, but there is a lot of hustle to it too. I think he would consider it a waste and not fun any other way.
    - panick21 5y ago
      No it isn't. That literally the dumbest thing I have read in this thread. So your friend is the CEO two of the most successful tech companies of the decade. Specially a person with the history of having the exact follow-through you criticize him for. Have you literally not been around for the last 2 decades? Funny that this person that you have diagnosed drug addiction is widely more successful then you are.
  - panick21 5y ago
    This is just such nonsense, sometimes I question if people ever actually adsorb information or just form lose association and then form an opinion based on how much they like the person involved. > Remember the Cybertruck? Cybertruck is literally not vapoware. It was never even supposed to come out before the end of this year. Technically they are not even late yet. Now they have pushed it back into 2022 but that is a few month delay is really not a that big a deal for a truck product with a new architecture based on a new factory. The way you talk is as if the Cybertruck was supposed to be out last year and was delayed to 2025. Literally every indication is that this will be a very successful product for Tesla. It being a few month, or half a year late is really not that big deal for a engineering and manufacturing project of that scale. > Or how he said he could control pigs by putting microchips in their brains? That was never what they actually said. Why the need to just make up stuff? And again, the never claimed this was a product anybody can just go out and buy within a year or so. Neurolink the company is still working on that and literally nobody expected them to have a product yet. Why do you feel the need to shit on the engineers that work on this just because you have some issue with Musk? The companies first goal is to make a product the help people with different deceases and disabilities. Do you get equally dismissive of all medical companies that don't have a product in the market after a few years of fundamental research? So just because you have made up a wild story about Neurolink being about controlling sheep and you don't remember seeing a video of a remote controlled sheep on twitter you conclude that it is all bullshit. > It’s probably just about doing whatever will impress Grimes for a little while. This sort of psychoanalysis is literally the lowest form of comment you can find on Hacker News. You should be ashamed of yourself. > Then they both smoke so much weed they forget the previous never-gonna-happen idea and come up with a brand new one. Yeah because a Musk company has never actually done what they said they would. Literally all Musk does is make up stuff and that is why his companies constantly fail and never produce any product. No matter SpaceX will launch 50-80% of total payload to Orbit this year. No matter that SpaceX produced the best rocket engine in human history. No matter SpaceX is universally accepted as the globally dominate space company. They control more satellites then all other government and companies combined. No matter that Tesla is universally acknowledged as the global leader in BEV, auto-selling the second best most prolific BEV company by 2x and volume and far more then that in revenue. No matter they produced the fastest production car in human history. No matter that they literally designed a new battery and battery factory from the ground up. No matter that both company are growing widely fast and Tesla being very profitable. SpaceX was very profitable but huge investment in Starship and Starlink likely make it unprofitable right now, but the unit economics are still great on their core business. But of course, these are clear signs that all he does is sit around and get high and comes up with ideas to impress his girlfriend. Honestly, I can not even imagine what level of dislike or jealousy somebody must have to be so dismissive of somebody else. If you want to call him a social inept asshole that is fine, but your argument simply doesn't hold water.
- mciancia 5y ago
  Why? I mean, what are current problems with making this work? Hardware/mechanical part? Software? Energy source? All of those? I would assume it's software, same as with level5 self driving (?)
  - tinco 5y ago
    No, robotics is not a software problem. If it was then it would actually be feasible. All it has to do to be useful is walk around without bumping into things, Boston dynamics already has the software for that.
    - yreg 5y ago
      What problems does Boston Dynamics have?
runawaybottle 5y ago
But "useful humanoid robot" is way more forward-looking than anything else he's done this with I think flipping a giant rocket on its way back down to have it land upright sounded just as nutty as this awhile back, let alone a startup competing (successfully) against the military industrial complex. The entire car industry is also adopting electric. No, I think Musk has enough notable wins to get a pass here. It seems as though a lot of people (not the GP) really want to be apart of the narrative change on Musk, whatever humanity’s obsession is with building up and then destroying someone, followed with the redemption story. He really doesn’t have to prove anything anymore in my book.
- piiswrong 5y ago
  But if you really think about it, flipping rocket is a trivial toy problem in simulation. But useful humanoid robot is very hard even in simulation. Just look at some of the SOTA control theory research. Their simulated robots looks stupid.
  - runawaybottle 5y ago
    Universal Healthcare is also a toy problem, but a true albatross concept beunricratically in the US. I say it’s a toy problem because we literally have copy and paste templates from most of the western world that we can use. Flipping that rocket and getting NASA contracts to go actually do it against Boeing/Lockheed/etc is not a toy problem. I don’t want to say it would never have happened without Spacex, but probably it would not have happened in our lifetime.
    - majewsky 5y ago
      Flipping rockets is a simpler problem in the sense that it the required technology already existed, there just wasn't anyone with enough incentive and resources to actually do it. Human-level robotics, on the other hand, is not a problem of missing incentive. For all I know about the field (as a casual observer), we are still several breakthroughs away from achieving this goal.
FireBeyond 5y ago
> If I had to make a bet on when it was possible, I'd say 25 years. Maybe 10 if Tesla goes all-in on robots and spends tens of billions on it, Manhattan project style. Except Musk says they'll be in beta in 2022. Hah.
- natch 5y ago
  >Hah. Crappy beta can be done quickly and can lead to other things. https://en.m.wikipedia.org/wiki/Falcon_9_prototypes https://en.m.wikipedia.org/wiki/Falcon_9_prototypes
croddin 5y ago
> But "useful humanoid robot" is way more forward-looking than anything else he's done this with Other aspirational goals he has set include a colony of a million people on Mars and a neural lace. I think a humanoid robot is less ambitious than either of these two.
- johnla 5y ago
  I feel like with Tesla navigational brain technology, they can easily put together a bot that can mow lawns, sweep and mop floors. Pull out the garbage/bring cans back in. Security Sentry observe. Get mail. Walk dog. Deliver packages. I would pay for a bot that takes those chores off my hands.
- hdjjhhvvhga 5y ago
  And, frankly speaking, it is a goal worth pursuing: not just per se, but for all technological advancements it would need to make in order to be successful.
- marticode 5y ago
  We already have the technology to send people to Mars. We sent people on round trips to the moon 50 years ago, so a one way ticket to Mars now isn't really that hard. What we don't have is cost effective technology to send enough people and equipment for a permanent colony there. But at this point it's really just working on getting cost down (which Musk has done beautifully so far), we have a pretty good idea of how we'll get there. On the other hand we just do not have the technology for a full humanoid robot, neither now nor the near future. An AI that can do mundane tasks like ironing and folding clothes is way beyond our grasp, and then there are the issue of limbs and power too.
  - inglor_cz 5y ago
    "We sent people on round trips to the moon 50 years ago, so a one way ticket to Mars now isn't really that hard." IDK, man, life support systems for such a long trip are much more demanding than for a jaunt to Moon and back.
notahacker 5y ago
"Useful humanoid robot" is easily pared down to "robot toy with a vacuum cleaner attached, and some funny gestures on top of an Alexa-style Internet querying system". And a lot of promises that this time next year it'll be useful, or at least able to climb stairs. That level of technical functionality is already achievable, and the Musk brand solves the commercial problem of "but it doesn't really do very much"

OnlineGladiator 5y ago

> If I had to make a bet on when it was possible, I'd say 25 years. Maybe 10 if Tesla goes all-in on robots and spends tens of billions on it, Manhattan project style. I have a decade of experience in robotics (including both humanoid robots and self-driving cars) and I think this is optimistic. Robotics is a wonderful field if you like to work on hard research problems. It is a terrible field if you're trying to invent a profitable product.

pontusrehula 5y ago
I think industrial robotics is a well established and profitable field.
- weego 5y ago
  Industrial robotics focuses on single task, highly specialised machines that have few constraints on form, size etc. This is not that. What tesla are conceptually pitching here (let's ignore that they'll deliver something that's pretty useless by comparison to what they're hoping you're assuming they mean) is a not-in-our-lifetimes leap in robotics in multiple ways
- squarefoot 5y ago
  True, but humanoid robots have little to do with industrial applications. I'm not bashing Tesla or Musk in any way, but this seems more of a mix of PR and research. Robotics is indeed the future, there will be huge demand in a few years, and of course they want to keep a foot in that door, so they need a project which today allows them to say "we're doing robotics too", tomorrow could evolve in big "ooohhh!" and in the meantime let them build experience in the field. In due time all money spent in this field is going to return. In my opinion this robot will probably evolve either into a companion aid for lonely/ill people or maybe a future Tesla luxury car chaffeur, rather than doing any serious mechanical work. For most uses, today a wheeled/tracked platform with grabbers is still cheaper, stronger, better. The knowledge accumulated during the journey to get there however will represent the most valuable asset.
- fragmede 5y ago
  that Tesla is in the process of building up expertise in. The Fremont, Nevada, China, and German Tesla factories are all filled to the brim with industrial robots.
- _trampeltier 5y ago
  That 100% something else. Industrial robots are almost allways in a cage for safety. I know there are 'cobots' now, but I just never saw one, except in showcases. Industrial robots just do a fixed task and are dumber than a slice bread. Industrial robots are super heavy for what they can lift. Even they are very efficient with energy, they are still allways on main grid / no batteries. The huge task is anyway not to build a robot. The huge part is to make the robot (or any device) smarter than a slice bread.
cush 5y ago
If solid state battery tech takes off, it's entirely possible within a decade.
- HWR_14 5y ago
  > If solid state battery tech takes off, it's entirely possible within a decade. Your contention is that its power consumption that makes a general purpose humanoid robot impossible?
  - cush 5y ago
    Yes. Great conversation.
tessierashpool 5y ago
Musk isn’t actually looking for profitable products. He’s trying to be the king of vaporware. Remember the Cybertruck? Or how he said he could control pigs by putting microchips in their brains? Or how Tesla was all in on Bitcoin until it wasn’t? He probably doesn’t even do it to bump the stock price. It’s probably just about doing whatever will impress Grimes for a little while. Then they both smoke so much weed they forget the previous never-gonna-happen idea and come up with a brand new one.
- tessierashpool 5y ago
  people can downvote this all they want, but if you’ve ever had a friend who smoked too much weed, or fallen into the habit yourself, you’ve seen the same behavior that Musk exhibits. too many ideas, not enough follow-through, and the ideas tend to be wild and unrealistic —- he’s a textbook case.
  - fitzie 5y ago
    Elon is willing to commit a lot of capital and effort into his ideas. the production capabilities of Tesla are improving at a significant clip, and incorporating many state of the art techniques. Elon's approach to innovation is a lot of hype, but there is a lot of hustle to it too. I think he would consider it a waste and not fun any other way.
  - panick21 5y ago
    No it isn't. That literally the dumbest thing I have read in this thread. So your friend is the CEO two of the most successful tech companies of the decade. Specially a person with the history of having the exact follow-through you criticize him for. Have you literally not been around for the last 2 decades? Funny that this person that you have diagnosed drug addiction is widely more successful then you are.
- panick21 5y ago
  This is just such nonsense, sometimes I question if people ever actually adsorb information or just form lose association and then form an opinion based on how much they like the person involved. > Remember the Cybertruck? Cybertruck is literally not vapoware. It was never even supposed to come out before the end of this year. Technically they are not even late yet. Now they have pushed it back into 2022 but that is a few month delay is really not a that big a deal for a truck product with a new architecture based on a new factory. The way you talk is as if the Cybertruck was supposed to be out last year and was delayed to 2025. Literally every indication is that this will be a very successful product for Tesla. It being a few month, or half a year late is really not that big deal for a engineering and manufacturing project of that scale. > Or how he said he could control pigs by putting microchips in their brains? That was never what they actually said. Why the need to just make up stuff? And again, the never claimed this was a product anybody can just go out and buy within a year or so. Neurolink the company is still working on that and literally nobody expected them to have a product yet. Why do you feel the need to shit on the engineers that work on this just because you have some issue with Musk? The companies first goal is to make a product the help people with different deceases and disabilities. Do you get equally dismissive of all medical companies that don't have a product in the market after a few years of fundamental research? So just because you have made up a wild story about Neurolink being about controlling sheep and you don't remember seeing a video of a remote controlled sheep on twitter you conclude that it is all bullshit. > It’s probably just about doing whatever will impress Grimes for a little while. This sort of psychoanalysis is literally the lowest form of comment you can find on Hacker News. You should be ashamed of yourself. > Then they both smoke so much weed they forget the previous never-gonna-happen idea and come up with a brand new one. Yeah because a Musk company has never actually done what they said they would. Literally all Musk does is make up stuff and that is why his companies constantly fail and never produce any product. No matter SpaceX will launch 50-80% of total payload to Orbit this year. No matter that SpaceX produced the best rocket engine in human history. No matter SpaceX is universally accepted as the globally dominate space company. They control more satellites then all other government and companies combined. No matter that Tesla is universally acknowledged as the global leader in BEV, auto-selling the second best most prolific BEV company by 2x and volume and far more then that in revenue. No matter they produced the fastest production car in human history. No matter that they literally designed a new battery and battery factory from the ground up. No matter that both company are growing widely fast and Tesla being very profitable. SpaceX was very profitable but huge investment in Starship and Starlink likely make it unprofitable right now, but the unit economics are still great on their core business. But of course, these are clear signs that all he does is sit around and get high and comes up with ideas to impress his girlfriend. Honestly, I can not even imagine what level of dislike or jealousy somebody must have to be so dismissive of somebody else. If you want to call him a social inept asshole that is fine, but your argument simply doesn't hold water.
mciancia 5y ago
Why? I mean, what are current problems with making this work? Hardware/mechanical part? Software? Energy source? All of those? I would assume it's software, same as with level5 self driving (?)
- tinco 5y ago
  No, robotics is not a software problem. If it was then it would actually be feasible. All it has to do to be useful is walk around without bumping into things, Boston dynamics already has the software for that.
  - yreg 5y ago
    What problems does Boston Dynamics have?

runawaybottle 5y ago

But "useful humanoid robot" is way more forward-looking than anything else he's done this with I think flipping a giant rocket on its way back down to have it land upright sounded just as nutty as this awhile back, let alone a startup competing (successfully) against the military industrial complex. The entire car industry is also adopting electric. No, I think Musk has enough notable wins to get a pass here. It seems as though a lot of people (not the GP) really want to be apart of the narrative change on Musk, whatever humanity’s obsession is with building up and then destroying someone, followed with the redemption story. He really doesn’t have to prove anything anymore in my book.

piiswrong 5y ago
But if you really think about it, flipping rocket is a trivial toy problem in simulation. But useful humanoid robot is very hard even in simulation. Just look at some of the SOTA control theory research. Their simulated robots looks stupid.
- runawaybottle 5y ago
  Universal Healthcare is also a toy problem, but a true albatross concept beunricratically in the US. I say it’s a toy problem because we literally have copy and paste templates from most of the western world that we can use. Flipping that rocket and getting NASA contracts to go actually do it against Boeing/Lockheed/etc is not a toy problem. I don’t want to say it would never have happened without Spacex, but probably it would not have happened in our lifetime.
  - majewsky 5y ago
    Flipping rockets is a simpler problem in the sense that it the required technology already existed, there just wasn't anyone with enough incentive and resources to actually do it. Human-level robotics, on the other hand, is not a problem of missing incentive. For all I know about the field (as a casual observer), we are still several breakthroughs away from achieving this goal.

FireBeyond 5y ago

> If I had to make a bet on when it was possible, I'd say 25 years. Maybe 10 if Tesla goes all-in on robots and spends tens of billions on it, Manhattan project style. Except Musk says they'll be in beta in 2022. Hah.

natch 5y ago
>Hah. Crappy beta can be done quickly and can lead to other things. https://en.m.wikipedia.org/wiki/Falcon_9_prototypes https://en.m.wikipedia.org/wiki/Falcon_9_prototypes

croddin 5y ago

> But "useful humanoid robot" is way more forward-looking than anything else he's done this with Other aspirational goals he has set include a colony of a million people on Mars and a neural lace. I think a humanoid robot is less ambitious than either of these two.

johnla 5y ago
I feel like with Tesla navigational brain technology, they can easily put together a bot that can mow lawns, sweep and mop floors. Pull out the garbage/bring cans back in. Security Sentry observe. Get mail. Walk dog. Deliver packages. I would pay for a bot that takes those chores off my hands.
hdjjhhvvhga 5y ago
And, frankly speaking, it is a goal worth pursuing: not just per se, but for all technological advancements it would need to make in order to be successful.
marticode 5y ago
We already have the technology to send people to Mars. We sent people on round trips to the moon 50 years ago, so a one way ticket to Mars now isn't really that hard. What we don't have is cost effective technology to send enough people and equipment for a permanent colony there. But at this point it's really just working on getting cost down (which Musk has done beautifully so far), we have a pretty good idea of how we'll get there. On the other hand we just do not have the technology for a full humanoid robot, neither now nor the near future. An AI that can do mundane tasks like ironing and folding clothes is way beyond our grasp, and then there are the issue of limbs and power too.
- inglor_cz 5y ago
  "We sent people on round trips to the moon 50 years ago, so a one way ticket to Mars now isn't really that hard." IDK, man, life support systems for such a long trip are much more demanding than for a jaunt to Moon and back.

notahacker 5y ago

"Useful humanoid robot" is easily pared down to "robot toy with a vacuum cleaner attached, and some funny gestures on top of an Alexa-style Internet querying system". And a lot of promises that this time next year it'll be useful, or at least able to climb stairs. That level of technical functionality is already achievable, and the Musk brand solves the commercial problem of "but it doesn't really do very much"