OPTIMUS Gen 2 w/ James Douma & Scott Walter. Diving Deep w/ Tesla AI & Robotics Experts PART 2
HTML-код
- Опубликовано: 4 фев 2024
- In this video, James Douma, Scott Walter, and I discuss the current capability of Optimus Gen 2, and the changes we expect to see in both software & hardware over time as Tesla creates a more and more capable humanoid robot.
This Part 2 of a 2 Part conversation. If you’d like to listen to Part 1, you can find it here:
• OPTIMUS Gen 2 w/ James...
Shopping at Tesla? You can support the channel & save by using my Referral code:
ts.la/hans71453
PANELIST INFO:
James Douma
X Profile - x.com/jamesdouma
RUclips - @JamesDouma
Dr. Scott Walter
X Profile - x.com/GoingBallistic5
CHANNEL LINKS:
Hans' X profile - x.com/HansCNelson
Hans' RUclips channel - @HansCNelson
MY BUDGET SETUP:
Laptop: M1 Macbook Air
Camera: iPhone SE (Continuity Camera)
Mic: Shure MV7
Audio/Video Interface: OBS
Streaming Platform: Streamyard
Want to create live streams like this? Check out StreamYard: streamyard.com/pal/d/60406730...
#tesla #elon #musk #elonmusk #teslaengineering #teslaai #teslarobot #teslabot #ai #aibot #aibots #optimus #engineering #robotics #humanoidrobot #deepmind
Tesla, Elon, Musk, Elon Musk, Tesla Engineering, Tesla AI, Tesla Manufacturing, Tesla Dojo, AI, AGI, Technology, Software, Computing, machine learnings, AI models, Engineering, Tesla Bot, Bot, Optimus, Optimus Gen 2, Tesla Optimus, James Douma, James Doma, Scott Walter, Dr. Scott Walter, GoingBallistic 
Развлечения
James is certainly a little more realistic in his discussions, while elsewhere I'm hearing a lot of hype and conjecture.
Thanks again Hans. Part 2 was as good as Part 1. As always, James and Scott, your discussion was fascinating.
I agree that telling a robot to make a cup of coffee with a place, close, and push coffee maker is not sophisticated. It can be done with a PC and existing software (e.g. Aloha) including fixing the pod placement. I hope that we missed something because we already have numerous robots performing more difficult and/or more complicated tasks.
For those who are new to the field here a just a few:
Amar-6 - Simple tasks including using a hand drill and a hammer
Apollo - Factory/warehouse lift and carry
Digit - Unloading trailers and moving packages
Eve - The first general robot to enter the workplace. Uses AI to learn to tasks and to improve performance.
Kime - Robotic bartender
Walker X - General robot that can walk, climb stairs, balance on one leg, serve tea, water flowers, use a vacuum
Tesla wasn't the first automobile manufacturer to enter the humanoid robot arena. Honda worked on an number of variations on the theme with the most notable being Asimo which was introduced in 2000 with the inconsequential price tag of $2.5 million US dollars.
For a general function humanoid robot, It makes the most sense for rapid task programming to marry a variation of today's sophisticated computer aided manufacturing (CAM) software with an LLM task generator that allows various forms of prompting and with a multisensor quality assurance "fault detector" AI.
I still feel that both of you are not adequately addressing the QA value of a robotic factory worker. Only a fraction of its work-cycle time will require the use of the numerous sensors. QA to a level unknown in today's manufacturing will be possible at minimal added cost once the bots are deployed in the factory.
I also think you two should discuss Tesla's probable Optimus skunk works at AI Headquarters in Palo Alto and Tesla's most likely candidate to achieve real world AGI. That candidate is not Optimus and it is not FSD. It is the new increasingly AI based training system that Tesla is using to address all aspects of AI training. It already has skills that as best I can tell no other AI system possesses. Tesla's new Training System is currently performing the following tasks:
1. Assists in fleet data acquisition and curating.
2. Autolabeling of acquired datasets including fixed 3D data and 4D ego extraction, compression, and storage.
3. Management of Tesla's ever expanding almost-live real-world roadway geofencing.
4. Data regeneration including introduction of environmental variations and alternative egos.
5. AI training supervision and reinforcement control.
6. Generation and analysis of futures projected from existing datasets
7. AI compute optimization including hardware allocation, data prefetch, routing, and storage, and sequential AI training optimization.
Integrating these task reduces compute, storage, and energy consumption while it accelerates AI development. To get the most out of this system, the obvious investment is to train the system to sequentially improve its capabilities. Given the size of the available datasets and the goals, the Tesla training system may have the best chance of any AI system to achieve real world AGI.
PS: Be sure to check out my other recent posts regarding Gen 2 in particular.
Hans 👍. James 👍. Scott 👍 Thanks
One of the more realistic discussions on the topic that I've seen, i.e. acknowledges unknowns vs. making assumptions and applies a modicum of scientific skepticism.
Thanks for bringing James and Scott again! Hope James will be around more! Awesome show, Hans. 👌🏽
Looking forward to, "Drive like Steve Macqueen."
I value my time my time was well spent watching this young man. The highlender PS kindness is always tried. Following you on Xi can't find you.😊
Fantastic content Hans! James and Walter exclamation point two of my favorite guests the granularity and signal-to-noise is so rich
Thanks for all the information and a great conversation!
Delightful respectful discussion among real experts exploring the near future! The scaling can be incremental - there will be lines to produce the actuators and other components which will be needed for any complete design will need production lines. Parallel production and scaling will be time efficient using different resources. Thank you all!
I loved his referenceto the pizza. Delivery on college campuses with the little ice box robots. Because I spend a fair amount of time in Bowling Green, Ohio. And these little guys are everywhere I got pictures. You'd really enjoy like 3 of them stuck in the snow. The Highlander😊
Great discussion , thank you all🙏🏼👏🏽👏🏽👏🏽
I always walk away Having learned something. Bravo
why walk away? lol most of us sitting down wwatching :)
James and Scott are great resources. Thanks for bring them together in a great panel.
Thanks to you all.
I see these robots better suited than humans for space travel.
Great job guys helping others to understand a new technology.
Well Done James - thank you .
Thanks great discussion.
One critical requirement for the dexterity of the hands is the opposing thumb action. We are the only primates that have this ability, allowing us to make and use tools. So it would make sense to invest time and resources in fine-tuning the thumb articulation?
FSD is constrained by the scope / scrutiny in public. Robots in factory have lesser hurdles since tasks to be done are the triple D that are more simple.
Excellent and insightful conversation.
Love Hans and particularly these guests!!
Chapter 1: Introduction and Overview
Timestamp: 0:00 - 0:37
Summary: The introduction sets the context for the conversation as part two of a deep dive into Tesla's Optimus Gen 2 humanoid robot. It is noted that this part can stand alone, providing value without the need to listen to part one.
Chapter 2: Optimus' Capabilities in a Factory Setting
Timestamp: 0:37 - 3:05
Summary: Discussion on the potential roles of Optimus in a factory. Focus is on its physical capabilities, specifically in tasks like feeding parts to traditional robot cells. Limitations in speed and movement are noted, as well as potential tasks it could handle in its current state.
Chapter 3: Software and Planning for Optimus
Timestamp: 3:05 - 9:15
Summary: Exploration of the software challenges and milestones for Optimus. Discussion on end-to-end neural networks for lower body control, planning actions based on objectives, and advancing the robot's capabilities through more sophisticated tasks. Emphasis on software flexibility and general task execution.
Chapter 4: Limitations and Future Hardware Modifications
Timestamp: 9:15 - 17:58
Summary: Examines the current hardware limitations of Optimus, such as its inability to play the piano or perform tasks requiring high dexterity. Discussion on its slow locomotion and limitations in flexibility, particularly in factory settings. Future hardware improvements are contemplated.
Chapter 5: AI Training and Software Development for Optimus
Timestamp: 17:58 - 24:51
Summary: Deliberation on the AI training methodologies for Optimus, including the use of demonstrations and planning breakthroughs. The complexity of training a robot to perform tasks based on human demonstrations and the role of large language models in enhancing robot capabilities are discussed.
Chapter 6: Scaling Manufacturing and Deployment of Optimus
Timestamp: 24:51 - 42:21
Summary: Conversational shift to manufacturing scalability and deployment of Optimus. Speculations on the number of robots needed for effective training and when Tesla might scale up production. The potential economic impact of deploying Optimus in factories is also explored.
Chapter 7: Comparison with FSD Development and Future Projections
Timestamp: 42:21 - 55:27
Summary: Comparing the development timelines of Tesla's Full Self-Driving (FSD) technology and Optimus. Discussion on the potential timeline for Optimus to perform useful tasks, both in factories and general-purpose settings. Speculation on how advancements in AI will influence the capabilities and deployment of humanoid robots.
Closing Remarks
Timestamp: 55:27 - 55:27
Summary: The video concludes with a thank you to the viewers and an encouragement to share the content with others interested in Tesla and its projects.
I think it is safe to say Optimus will never be ready. Just as a human will never be ready. But it will be more and more ready and be able to perform more and more advanced tasks and therefore be more and more useful everywhere.
Ready for what?
@@michaelholmes8848Ready to for everything or whatever James and Scott meant.
@@perjohanohlsson so it will never be ready for basic tasks on the vehicle production line?
@@michaelholmes8848 Yes, it will. What I meant is there will not be a time when we can say it is ready in general buy ready for certain tasks. Just like FSD will never be ready but many times safer than humans. "March of 9s"
Cheers guys
“FSD as a copilot “. that has been my experience exactly having been working with FSD for over two years. I am both comfortable with its limitations and ready to intervene should it surprise me. But it absolutely makes my driving experience much more relaxing and I too am certain that it makes me a safer driver even though it is still not SAE level 3 yet.
Hans please get your Patreon started!
Thanks for the reminder!
I've been meaning to do this for a while now.
Just got the basics set up.
You should be able to find it on Patreon under HansCNelson.
I don't want to post the link here since I don't want to create an opportunity for scammers.
@@HansCNelson I just checked but don’t see it yet, I’ll give it a day and see if it’s published and let you know here if I can see it.
Looks like you found it. Thank you so much!
@@HansCNelson You’re welcome!
41:22 it appears that Tesla may go into the electric actuator business.
I can't load the dishwasher to please anyone but myself. So we have to not just get the robot working we have to get people to accept the robot and it's work. The hardest part of the equation is understand is public acceptance. That's a great part of Tesla's advantage is that the whole development and use case will be in-house to start with. Tesla has the technical chops to design, build, train, and use a massive number of robots without having to make it practical or even cost effective in the beginning. That's unique to Tesla and will be for a good while.
Every person will take over the strength to rise above and follow the three Laws
18:15 Re: Cloths Folding Needn't be Slow. The main deterrent to rapid clothes folding is aerodynamics. Once a neural networked bot learns the process of generically folding clothing, e.g., along "hang" lines, if placed in a vacuum & trained to determine the inertial & gravimetric effects of force, direction, & speed imparted at each movement on the materials being manipulated, bots will make quick work of clothes delivered from whatever cleaning processes.
I would also think that they will also ask vendors to use the bots as well. This would help reduce the product costs in a considerable way.
That would make a lot of sense
Should the robots have the equivalent of a PTO? A plug to connect to control a power drill for example?
Thanks, Hans (and James and Scott), great conversation. The Future is rapidly approaching.
it is interesting to consider the (huge) difference between "training" a neural net model and then applying that model to control a body to perform a task (more or less what the human autonomic nervous system does) and, on the other hand, arranging/making it possible, for the robot to "learn" to perform a task (competently and efficiently) and improve on that learning through practice and dynamic feedbacks. In the latter case, the neural net needs to be "trained" literally to "learn". This seems to be what James and Scott are calling "planning" but that also requires understanding the goal and maybe even the purpose.
FSD models are trained to appropriately deal with millions of data cases and then they can also handle many, many data cases that are similar but never seen before. But once trained, the model is "fixed" and does not learn or change. (Subnets might observe the driver and state of the car, etc. and adjust some parameters, seeming to "learn", but likely that variability is "contained" in the training data too.). If it turns out in certain situations, model performance is sub-optimal, then the training datasets can be extended to cover those and the FSD model re-trained and cars updated.
Human autonomic nervous systems learn very slowly and do change from infancy to adulthood to older age, from genetic reflexes into sophisticated psychomotor "skills". There is "plasticity." But people (and animals, maybe even plants) do learn rapidly and "consciously," through awareness of themselves as embodied agents within some situational context and with given or imagined goals and purposes. Then "planning" becomes a hypothetical cause-effect process series to get from a state of "now" to a state of "then." The real-world implementation of the action-plan is "behavior." All this requires and is facilitated both by knowledge (long term memories) and by the ability to recognize and deal with all of the known or unknown barriers between now and then. That is why "learning" requires perception, feedback, and practice - i.e., "experience." It is why "teaching" (particularly through demonstration and/or reward) can help and why optimization and competency can "develop."
Embodied neural net models can be trained to change/evolve/learn from stimuli rather than simply react; to develop and modify action plans based on given or generated goals; especially if provided realistic world models and good situational/perceptual contexts. This is "nurture/nature" iteration and an important goal for AGI.
Ha, apology for such a long comment! Spent too long thinking about what I wanted to say. Anyway, Tesla is obviously making amazing progress on all fronts: autonomic balance and bodily control as well as action planning and agency.
Looking at today’s assembly line is not the correct way to think about this . The unboxed process has been designed FOR Optimus & all its current constraints . 4 Down the Chess board .
Shorter questions are better 😁
I worked my way through college doing dull, dangerous and dirty jobs. Optimus is already close to being able to take over some of those jobs I had
Ditto. I recall installing 1,280 rod bearings per hour for 8 hours. Longer if we were running overtime.
It seams to me, to be the most effective at a task, use an end of wrist effectors.
Optimus could enter a motor vehicle quite easily....if the seat was rotated toward the entry to 'greet' the user.
How did Kuka Robotics scale, or do we have a more similar example?
Ballet, first position.
Unlike a human, a bot could be attached to some large arm that could position the bot in odd locations.
You can't really grab a human and then turn them sideways and then poke them in a small space but one would think that one could do that to a bot.
So the arm could grasp a bot along the waste, lift it up and turn it to any orientation where it is suspended over the floor and then insert it into a tight location so that it does not have to bend all that much to access the area where it is to perform some task.
The Bot insertion arm could itself be controlled by the Bot.
For example one might Google
ST Robotics R12-six small low cost 6-axis benchtop robot arm.
and image an arm like that scaled up so that it could hold, lift , and orientate a humanoid robot.
Thus it could position and orientate a robot to access hard to access areas where the Bot might not need to do any strange contortions to access those areas.
Thus the Bot could do something that a human could not (or would not) do, that is being (repeatedly) bodily grabbed, lifted, and positioned in strange orientations.
What a piece of work is man...how infinite in faculty. We're becoming quite aware of that now.
I am 41 and I wish the house maid robots can do basic house chores before I get old.
Hello guys
Your channel is my go-to when I need a pick-me-up. Thank you for that!
Why not marry an Optimus torso with an SPMT cart? The SPMT can rotate in any direction and carry heavy loads. The Optimus torso could then lift and transport items on the cart. Maybe even have a torso at both ends of the SPMT to handle heavy or large objects. The torsos could even be telescopically elevated (or the cart top scissored upward) in order to place objects on higher shelving or to other tasks. The SPMT cart could house a Tesla skateboard battery base to provide the power needs for both the cart and the torsos. I imagine additional arms could be added to the torsos.
We are in the midst of a Cambrian Explosion of robots.
The Bot will have to get in and out of a car. How else will you take it on a date??
Definitely needs more actuators and a nice skin before it will look good in a dress.
Wonder if they will be able to make both like surrogate movie with Bruce Willis 😅
LLM and Image Model alreay can check the states of real world situtaions and just left is the Robot Action Model and if for certain tasks you can use some things Nivida's minecraft preject concept, and as you want the robot to do as much tasks as them can you just need their body and moter to be design as hunman as much as possible
New ATLAS video is available.
just a terrific discussion. Thanks for setting this up. They touched on so many interesting points that I’ve been thinking about as well.
consider building the $25,000 car from a design for manufacturing standpoint. The manufacturing experts and the design team at Tesla can fine tune the design and the manufacturing line to accommodate and integrate the use of humanoid robots. This makes Tesla along with the $25,000 car a terrific case study for highly efficient manufacturing.
Musk is very forward thinking and I’m highly confident that this is the approach they’re taking. The legacy automotive companies need to be paying very close attention.
It will be a long time until Optimus will be able to walk my 100 lb. German Shepard. 🍌
That’s a tall task.
Literally 😂
Will replace Tesla factory workers at least a couple of years before they are ready to ship for general factory work. Years more before they can send them to end users... and that's not a problem as the replacement of their Own factory workers will be massively profitable.
Hans, James, and Scott, Here is an update on on where neuroprosthetics are today. It helps put Neuralink and Optimus in better historical perspective. ruclips.net/video/F_brnKz_2tI/видео.htmlsi=ANSm0ooEsvDKvewm
❤😊
Robot army would be good
To come after you, right?
@@flickwtchr if so
If it can’t jump, can it avoid falling if unexpectedly bumped (by human)? Hopefully speed to move in those situations is only a software constraint. Musk set a limit for safety reasons. Bot needs a burst speed exception on rare occasions, maybe.
That would probably be a good idea.
To me robots will be useful earlier than FSD, because of responsibility; cars can kill somebody and the driver is liable; a robot can drop a bolt or cut too much fabric, depending on the task one gives him. Risk management 101; robots will come soon.
How far along? When will they go into production? 24 25?
Everything learned on the road to v12 has been learned. Optimus stands on those shoulders. I expect this to be so quick it blindsides everybody except those within the dev team.
Tesla will give away the Optimus bots by 2030. The money is in the software subscription, not in the one-time sale of the bot! Do the math, it's very clear. The cumulative # of bots deployed, eventually reaching into the millions, tens of million, will save factory owners $36/hr (human worker + benefits). Therefore, the factory owner will be able to pay Tesla several dollars per hour to use the humanoids. Humanoids will be able to work 4x-5x as many hours/yr. as people. And, they will be more productive. Do the math of tens of millions of (total) deployed Tesla bots x $3/hr x 7,000 hrs per year. That revenue will be like 75%+ margin. The $36/hr. will be split between Tesla, the factory owner and the Gov't (new taxes to possibly fund universal income)...
(look what i can do )
Societal consequences when it moves out into the labor pool…uh, yeah, you mean like a LOT of human beings losing the means to feed their kids, and stuff like that? I know, I know, those jobs are boring, tedious, and dangerous so being homeless on the street is a much better outcome for the human. But hey, gotta get a piece of that multi-trillion USD pie coming from what will be the largest transfer of wealth in the shortest amount of time in the history of civilization. But then, we are to believe that libertarian economic types, or neoliberal economic types are going to be leading the charge to distribute to the little people a universal basic income, which is a fantastical notion. And if such an income will be distributed at what cost to the freedom of the individual? Biometric tracking of the human? Real time tracking of expenditures allowed or not? Social credit scoring determining availability of income to begin with, or qualification for increase? Can't you just smell the freedom of all of it?
how unimaginative must folk be to lmit somethng to 2 arms, 2 legs and one head zzzzzzzzzz @:)
8:55 I would not agree, even humans must get training to work even for simplest job. So why we except to robots be better from start? I think they should aim as progressively more complicated goals which are jobs in factory. Because aiming at universal robot, it may be too late to use it, and people working on it don't focus on important think: making it do job.
So goal is to show it once, and it do job, even when some things are not at same place, because this is problem with today factory robots.
11:25 about general assembly, IMO it was how unboxing method was created, they think how robots can do this job, and they invent this way. So $25k Tesla will be "Tesla Bot ready"
15:30 in part 1 was show that using simply robot with only 2 "fingers" can do a lot. I think James overthinking stuff, and I don't know why focus on what it can not do, and not on what it can do?
38:20 please, ask 1 question, because it was interesting what timeline they think, but because you ask too many at once they answer some 2040 predictions.
The chances of a 3rd party reaching AGI that can drive a robot/car better than FSD are not zero. if that's the case is TSLA still a good investment?
But how without the data and training?
No reason I can discern why you can’t improve an autopilot program using simulations.
@singed8853 I don't know, I am.not an expert. The tesla approach however seems to me more pragmatic. Imagine nvdia would be today where tesla is with their data and the vehicles. Wall Street would surely valuate this highly. However it doesn't do so from tesla and partially they are right as it was overpromised and under delivered till now.
Saying that a probability is not zero is about as information-free a statement as you can make. All the important differences are beyond that number. 0.000000000001 and .9999999999999
How did 16 year old you learn to drive?@@diogenez471
You should buy and hold tech stock Crypto investment
There are many interesting stocks in many industries that you might follow. You don't have to act on every forecast, so I'll suggest that you work with a financial advisor who can help you choose the best times to purchase and sell the shares or ETFs you want to acquire
I'm an advocate for having a diverse investment strategy. I grew to a 6 figure mark with my portfolio having exposure to different areas of the market, including small and large-caps of the ETF index, blue chip stocks, coins, grade bonds and alternatives like cryptocurrency markets, as this helps manage the overall risk on my portfolio managed by my FA.. Credits to my adviser, Jasper, i have no doubt investing more
This is impressive how could someone go about getting investment guidance from a coach like that, would you mind sharing your coach~info?
He's always act~ive on Whats~App...🎉
They are called androids, not humanoids.
This is ridiculous. In the future, if I want to hire a nail technician, would I get a clumsy Optimus or the SmartNailTech3000? It has tiny grippers to hold fingers and tiny filers/scrapers/painters. It trims, cleans, paints, applies and dries in a fraction of the time and takes up almost no space. Just plug it in no batteries needed. Oh, but it can't sweep the floors. Should Optimus push a broom and pick up trash or should I let my new SmartRoomba3000 do it? Oh, but that can't paint the walls, should we watch how fast a humanoid climbing a ladder can paint compared to SmartPainter3000 that you can rent?
Besides the completely unproven AI, look at the fingers on Optimus! Could it even pick up a needle to thread it, start a roll of tape, put a lid on a cup, bag items, make change out of a register? How long would it take to peel a label off the backing, tie a shoe, pick up a pen and write? A person, without looking, can pick up just one small screw out of a pile and orient it in their hand. Please tell me the "easy" jobs a humanoid will be able to do! There are no "easy" jobs that aren't already automated. I'm an automation engineer, why don't they put AI on all the existing dumb robots. You could test and prove learning concepts/systems on existing equipment. There are all kinds of robots, controllers, vision systems etc. that could use a brain NOW.
Elon is wrong about the viability of a humanoid robots. Making it more human-like doesn't make it more useful, in fact now it's limited to what a humanoid can do and will NEVER do everything a person could. And making them all the same is worse! Can it lift heavy objects? Should it stand in a factory packing widgets and never use its legs? The factory fork-lift of the future will be driven by AI and have cameras to see all around. It won't have a steering wheel and wait for a robot to climb in and drive. You won't be maintaining hand/foot controls on the truck, or wasting time, or wearing out the robot and wasting the energy. The best part is no part! "It's made human-like so it can work in environments made for humans" Think about that. That is the antithesis of "First Principles" It compounds costs/complexity and reduces efficiency.
Enthusiasts will say "It's not like FSD because you don't have to be 100% accurate". Then it will fail in its job. If it drops a box of screws, breaks something expensive, falls and breaks itself, stops production etc. it's costs could be more than the savings. Enthusiasts will say "It doesn't need to be fast because it can work long hours". Then it will fail in its job. Factories are built for people to operate, so it will have to operate like a person. Asset utilization is important. Otherwise you'll have to create dumb jobs for it like overnight stock-boy. Call me when they demonstrate a stock-boy, if I'm not dead. Enthusiasts will say "we'll need them on mars" On mars you will want a smart crane/robot that can manipulate and join large habitat panels and move machinery. And you would want a smaller smart robot with tools instead of hands so it can manipulate/repair small cables/connectors quickly without having to pick up bulky hand tools. And it should have extending legs rather than get on a ladder. Wouldn't it be ridiculous to have humanoids operate cranes, use wire cutters and climb ladders? You may only need two humanoids to set up a base because humans didn't evolve to use technology.
The fud about the car was easy to ignore because the car was on the road. But the uncertainty and doubt about a humanoid bot that doesn't even have a brain yet is very real.
Optimus can't even do what the darpa challenge bots did 8 years ago.
See a bot untie a knot watch YT "why we don't have robot butlers yet"
see one load a dishwasher "Princeton Robotics - Russ Tedrake"
You've made this argument before, but it seems to me that in the sense you describe, everything - people, robots, etc. - always eventually or in odd circumstances .. fail. So what? Not nihilism. Instead, you get back up again and go on. Surely robots will be handy in agriculture (as so many itinerant workers are). That's 2 million jobs in America and many more elsewhere. And helpful in dangerous/hazardous working conditions - in factories, underground mining, space, etc., or disaster situations. Yes, a specialized machine can be designed for just about every job and every niche ... but given that the human world (as opposed to Mother Nature) is deliberately designed for the humanoid form factor, why not fit the robot to that world. Let it begin by doing boring things people don't like doing. I do see and, I think, understand many of the points you make ... and they have some validity. But, again, so what? Humanoid robots make sense in a human world and economic sense in many situations. And, to the extent they are AI or AI extensions, they should be treated well.
Screen capture this comment for future laughs.
@@WarrenLacefield " Humanoid robots make sense in a human world and economic sense in many situations."
Please describe one.
Agriculture? doing what? Wouldn't it be silly for a humanoid to work in a field and go back to charge itself? Wouldn't a wheeled robot with eyes and grippers work just as well, yet be less way complicated and way more efficient? aren't they already making picking robots?
Yes, I think you are missing the point. With APP stores this can have all of SmartNailtech ability. This is the base model of the first Apple iPhone. Would you compare the original with the iPhone 15 in speed and agility? Nope. They have to be able to fill certain tasks for the factory floor. Learn to walk before you run. 😊
@@davidvincent5244 You are missing my point. Everyone in making highly unlikely assumptions when the evidence points to the contrary.