Tesla profit soars. And also craters

Of course, any space also used by humans will be designed with legs as the expected transport mechanism unless the users are disabled. E.g., stairs … which are complicated to do with wheels.

Sure…but we’ve been living with the Americans with Disabilities Act for three and a half decades now. A very large number of workplaces are going to be completely accessible by wheels. And most factories, warehouses, and other industrial/assembly/storage places are going to be fully accessible by wheels without stairways simply because that’s how things are moved around already - even without accessibility requirements.

Yes, there are exceptions, of course. There are older facilities where some or all of the workspaces are accessible solely by stairs. But I expect those are the minority of circumstances.

I would be shocked if there was a single job in Tesla’s U.S. operations that couldn’t be reached by a wheeled robot (or a person in a wheelchair) without having to navigate stairs of any kind.

Not terribly relevant. Anyplace that requires tools or materials of any sort to be moved around is going to have elevators or lifts. Nobody is carrying car frames, medical supplies, or even brooms and mops up and down stairs. Legs may work for stairs, but we have designed the world to work with other things - except for us.

@Goofyhoofy @btresist

The unsolved problem will AI actually get things done correctly across a larger database? The jury is still out.

There is not a lot to proof that will happen.

My nephew(MIT programmer)'s company has an AI component in speech recognition. Five years ago he was told it would get past 99% accuracy. Two years ago perhaps it will get to 97% accuracy. That sucks in a lot of areas of life and work. I get humans fouled up. But daily making errors as a machine is different. The human translators and speech transcribers have better than 99% accuracy. There are a lot of small things in human speech patterns that leave AI coldly behind.

That said I am so bad in English that I use Grammarly which makes me a much better writer and slowly trains me to write more effectively.

Actually, observations that FSD works most of the time and that a number of other companies are already field testing their versions of autonomous driving strongly suggests that general purpose humanoid robots are coming soon. Driving is very complicated with little tolerance for error. In contrast, housekeeping tasks are relatively simple and does not have to be 99.99% accurate as no one dies from a poorly sheeted bed.

For example, one human could supervise multiple housekeeping bots and fix any errors that occur.

Why would the cost be high? Again, I suspect the processing power to drive a car under real world conditions is much greater than most low skilled work tasks. Once the learning algorithms and methodology are optimized, why would the robot cost more than a Model 3?

Once one has a general purpose humanoid robot capable of using human tools the need for specialize machines is much reduced. If my bot can use a power drill, leaf blower, and vacuum, why would I need separate specialized machines for these tasks? This allows Tesla to mass produce Optimus in a single form factor and market it for a variety of tasks. Economies of scale.

These conclusions are based on what, a 20th century view of technology?

I find it remarkable that folks with no technical expertise can make such conclusive statements about technical matters. At least three humanoid robot types will be field tested in 2024 (at Amazon, BMW, and Tesla) and China has already announced plans to mass produce such robots in 2025. I guess you can dismiss these efforts as that of liars and scoundrels, but it is possible they know more about what their products can do than you do.

I don’t think robotic bipedalism is a problem based on the demo videos from multiple companies. As with cars, it is all about the software and the capacity for mass production.

Here is a video comparing four humanoid robots. It is from youtube, so take it for what it is worth but it has some interesting info.

1 Like

I think talking about the impact of humanoid robotics with people that are invested in index funds is like talking about swimming the English channel with someone who has a life jacket on. The people with the life jacket really haven’t any understanding of the risks involved nor do they really care. In other words, if you do not have skin in the game you really could care less on the outcome.

Andy
Who realizes some people would only like to argue, not to better understand the risk, but to hear their own voice.

1 Like

I’m sure they do, but even they don’t think their products can do anything close to what you’re suggesting. As noted by the Tesla engineers in the last earnings call, they’ve got the robot to where it can walk around and do some things, but they’re still trying to get it to the point where it can do something useful - the utility is still the barrier. They’ll start throwing them into warehouses and the BMW/Tesla plants, because these things have an enormous halo effect (not scoundrels, just a good way to get tech people interested in working for your company). But the tech is still very far away from being able to do anything worthwhile.

It’s not about solving bipedalism - it’s about wasting that much of the robot on bipedalism. In nearly all situations, there’s no need for the lower half of the robot to consist of legs. If you use a humanoid form, nothing below the “shoulders” is useful for doing anything. As Goofy pointed out, any industrial or manufacturing facility will be accessible by wheels. My point about the ADA is that nearly every other type of commercial establishment (retail, service, hospitality) that’s been built or renovated in the last several decades is going to be accessible by wheels as well. A robot doesn’t need legs in most circumstances, except as an aesthetic choice, which makes all of that stuff wasted space and tech.

3 Likes

Standing, walking, keeping your balance requires a major amount of brainpower. Your brain is receiving multiple simultaneous inputs: eyes, Eustachian tubes in your ears, inputs from muscles like toes, feet, calves, thighs, and more, including where your arms and upper torso go to retain the upright posture. It why we are more prone to falling as we age: the inputs slow, the brain reactions and ability to simultaneously process so much slows - and we stumble and fall more often (and drop things, too.)

**Scientists and anthropologists believe that a human’s ability to stand on two feet requires a major amount of brainpower.** Balance and walking is an “all systems go” accomplishment. As you age, balance becomes more challenging and more important…especially for the health of your brain.

Your eyes, ears, brain and your sense of body position are the three main sensory circuits that work together to keep you in balance. Your eyes analyze ongoing visual cues and send signals to our brain to alert joints and muscles how to move when. At the same time, a tiny, fluid-filled tube in your ear canal sends messages to your central nervous system about the position of your head. All of these elements are controlled by your brain. So, if you’re experiencing loss of motor coordination, such as having a hard time balancing, it could suggest damage in the brain.

Why waste that energy and all the electromechanical requirements on something that is so much more easily achieved with “wheels”? (Because people are imitative, and it seems cool.) Also: wheels don’t exist as a locomotion device in nature, we had to invent them, and some societies never did.

(As albaby notes, a simple lower half tin can with wheels offers many advantages: energy/battery storage, tool storage, easier balance, better locomotion, fewer parts, lighter weight, among others. The only thing it doesn’t offer is the ability to disco, but having seen R2D2 shake a move I question even that.)

It’s not that bi-pedalism can’t be solved, it’s why bother? It’s expensive, it requires more parts which can fail, more energy to run, and offers almost no advantages (outside the animal kingdom.)

2 Likes

Which is irrelevant to me because I am talking about future prospects. If this discussion is limited to today, then I agree with you and Goofy, that based on public info there is currently no marketable humanoid robot. Where we disagree is that I believe there could and very likely will be one in a decade.

GM and Ford felt the same way about the marketability of EVs15 years ago. Because of that they are in serious danger of becoming niche companies only making trucks for Americans.

Again, that’s just your bias. You’ve presented no evidence favoring a “halo effect” over a genuine testing of what Musk calls “useful work”.

I am skeptical of your claim that all these companies are spending hundreds of millions and enormous intellectual resources on simply a recruiting tool. They could just offer more salary. I suspect they see a marketable product at the end of the rainbow.

Again, you considering it a waste of time is your bias. A general purpose robot has added value if it is able to climb stairs, ladders, navigate through uneven terrain, and step over obstacles.

But let me be clear, I don’t doubt there will be applications for wheeled robots or for those with more than two legs. I am not an ideologue on the issue. But I believe the basic form factor will be fundamentally humanoid, like Rosie the Robot on the Jetsons (showing my age)

With technology like neuralink capable of linking the brain to computers and leg actuators that can mimic human standing and walking, the necessity for the ADA could be temporary.

There are a lot of companies of the smaller variety where space in their facilities is more limited than a vast Amazon warehouse.

1 Like

No, that won’t work. Once a high-tier engineer reaches a certain salary level, and high-tier ones almost always reach that level quickly, a little more salary won’t entice then. Only work that specifically interests them will entice them to consider changing jobs (even within the company).

When you heard stories (they were true) a few years ago about companies like Facebook luring new employees with absurdly high salaries, they were just doing it to accomplish their “land grab” goals of hiring the extra 15,000 or 20,000 or whatever they decided they needed at the time. They weren’t getting all high-tier engineers, they were getting mostly mid-tier and maybe a few high-tier among them, especially if the person happened to be interested in the thing they would be working on. I think the oil and gas industry had a similar “land grab” effect a few decades ago, and they also paid top dollar for mid-tier staff, just to get them on the payroll.

2 Likes

I’m sure they do, as a best-case possibility. Just like Google expected to have marketable products from all their moonshot programs from Google X (Glass, 'Loon, Wing, Makani) and Faceboo…I mean, Meta, expected to have tons of marketable metaverse products from its tens of billions of dollars of investments. And just like Musk saw robotaxi fleets and a working Tesla Network by 2020. Doesn’t always work out.

But I’m sure these shops see a marketable product at the end of the road, even if it isn’t a humanoid robot. Humanoid robots are a great way of attracting personnel and funding (which is why BD developed Atlas), even if your company ends up selling other types of robots and not humanoid ones (which is why BD has abandoned Atlas as a commercial product and is only an R&D project now).

For Tesla (as opposed to all the independent shops), having a second AI product is a necessity. Sure, they could offer more salary. But realistically, would you expect to be able to get top talent in the AI field if you have only one AI project and it was almost done? Tesla’s a car and energy company, not an AI start-up. Before Optimus, the only thing Tesla needed AI for was the FSD project - and they were telling everyone all the time that they were a year or two from having that pretty much done.

And if we’re being honest, there’s a small part of me that thinks the only reason Musk is having Tesla do a humanoid robot is because he wants to use them to colonize Mars, not because he thinks they will generate value accretive to Tesla as a company. But that’s just the skeptic in me.

Thanks for the insight, though perhaps the folks being recruited are new MIT grads with the latest academic research background.

As I understand it, albaby’s premise is that the development of robotic bipedalism specifically and humanoid robots more generally is mostly PR for recruiting. Does that seem plausible to you?

It seems to me that the folks best able to discern relevant and marketable research from those of the “halo” PR variety would be the very engineers being recruited. They would know better than folks here whether they were being conned so it doesn’t seem like a very good strategy if albaby is right. I suspect young MIT geniuses still find stock options in successful companies attractive.

1 Like

So you say. On the other hand, here is Amazon’s explanation for investing in Agility Robotics and field-testing its humanoid robot, Digit.

“When we announced our most recent version of Digit earlier this year, this is exactly the type of repetitive material handling deployment we had in mind; one that enables humans to be more human.“Digit’s size and shape are well-suited for buildings that are designed for humans, and we believe that there is a big opportunity to scale a mobile manipulator solution,”” according to Emily Vetterick, Amazon Director of Engineering. “Collaborative robotics solutions like Digit support workplace safety and help Amazon deliver to customers faster, while creating new opportunities and career paths for our employees.” Digit is designed from the ground up to go where people go and do useful work, safely, in spaces designed for people. Because so many tasks are designed around human workflows, Digit’s human-centric design enables multi-purpose utility. Initial applications include bulk material handling within warehouses and distribution centers. Customers in the Agility Partner Program can expect delivery of the first Digits in 2024, with general market availability in 2025.

Amazon Begins Testing Humanoid Robots at Warehouses | ASSEMBLY.

I dunno, Amazon sounds like it is taking this seriously, but what do I know about robotics? I’m not a lawyer after all.

No, it’s not plausible. Much more likely the reason they are working on bipedal humanoid robots is because Elon Musk thinks it’s cool. And it’s is indeed partially for PR purposes, but not PR for recruiting specifically rather PR in general. Because a large portion of the US public grew up with robots on TV looking vaguely humanoid, that’s what they expect decades later in real life.

Whether or not there is an advantage for bipedal robots versus some other method of locomotion remains to be seen. Humans evolved over the last 5-10 million years to stand on two legs for some reason, but it doesn’t follow that robots need to have the same structure and method of locomotion. Sure, if you want the robot to sit in the drivers seat of a vehicle, then it needs to be so, but to change sheets? Maybe not. The beauty of robots is that they don’t need millions of years to evolve. We can build a robot today that is bipedal, but if that doesn’t work out well enough, we can build one with wheels in a year or five. We can keep both ro we can discard one if it doesn’t prove to be valuable enough.

I don’t think so. Stock options in a company that is already successful are rarely worth anywhere close to stock options in a company that is new and not successful yet, but then becomes successful (aka “a startup”). The trick is choosing which company will become successful. That’s not easy to do, and might explain why the “young MIT geniuses” tend to change jobs every year or two … because they are searching for the next big success story. But I wasn’t referring to engineers straight out of school, I was referring to those who already have reached the high-tier status. They are already paid well, they get their regular stock options, good working conditions, etc. But they will tend to work only on things that interest them. This is for two reasons, one, because they can, and two, because it is clear to them that their high-tier performance can only occur when they are working on something that interests them.

1 Like

This is what I’ve been thinking as a ‘lesser case’ scenario.
They might not get a profitable HUMANOID robot, but they are likely to get a ‘general’ purpose robot that can profitably perform 80% of the various tasks that a human would need to do.

Shoot for the Stars (humanoid form and functionality) and if you hit the moon (a profitable but non-humanoid, not quite as flexible/capable as a human)… is still a WIN!

:slight_smile:
ralph says thanks for confirming my bias

It’s not just PR. I think they’re serving the same role as “concept cars” in the automotive industry. They draw PR, true - and they are exciting projects for engineers to work on. They also provide opportunities to experiment and push the boundaries of the core product, even if they never themselves become commercially available products.

That’s why Boston Dynamics is continuing to work on Atlas, even though they are no longer developing Atlas as a commercial product. It’s an R&D project. Which is valuable to a robotics company, even though it’s not expected to result in a consumer being able to buy a humanoid robot from BD.

For Tesla, which was not previously an AI or robotics company, Optimus was valuable to provide a potential project for all their FSD people to shift into if/when FSD is finished. If all your company is working on in the AI space is FSD, and FSD is expected to be pretty much done within the year, it’s hard to offer people a viable long-term career path with your company.

How much do you suppose Ford sank into designing, manufacturing and marketing the Edsel? How about New Coke? How many gazillion dollars went into 3D TV? Google Glass? The McDLT? Smartphones from Amazon or Microsoft or Facebook? FTX? Metaverse? The Zune? Delorean? Segue? Pinto? Betamax? Lisa? Laserdiscs? Smokeless cigarettes? US Football League? Google+? WebTV? Not to mention “John Carter”, “Heaven’s Gate”, and “Cats.”

You seem to equate “spending a lot of money” to “this has to be a good idea”. Quite often, in fact, it’s not. And wouldn’t those companies have been better off if someone had stood outside the conference room and said, “You know, you might want to reconsider….”

1 Like

I agree that the economic case for a profitable humanoid robot doesn’t exist now nor in the next few years.
But the ability for a robot to self balance has been around for ~25 years. And thanks to Moore’s Law it is VERY cheap to do it.

Less than $20 at Walmart

The computing part is trivially cheap. And the sensing part is also cheap because it is basically the same sensors that are in every cell phone.

Somehow a bird’s brain can allow them to balance while sleeping on a perch, some even on one leg.

Mike

1 Like

Not at all. I’m saying that if lots of smart folks with a track record of success are investing a ton of money developing an idea and when major companies like Amazon, BMW, and Tesla are field-testing that idea, then that idea probably shouldn’t be so quickly dismissed as some are doing here.

Particularly by those who, from what I can see, have very little expertise on the topic.

But these days strong convictions don’t seem to require expertise. A brave new world.

Wheels do not go upstairs or crouch to get under things.

The parts will get cheaper and cheaper. This technology will be deflationary. Mass manufacturing will do that.