The Optimus robots at Tesla's Cybercab event were humans in disguise
theverge.com
theverge.com
I guess we can haggle over the definition of "robot". But they are humanoid machines that happen to be mostly remote-controlled - with some automation for functions like walking. If these aren't robots then neither are Boston Dynamic's dogs, and I have never seen anyone complain about people describing them as robots
Another example of this behaviour is Tesla themselves 3 years ago.
https://www.drive.com.au/news/tesla-mocked-after-unveiling-a...
You could also actually watch the event and listen to how Elon exaggerates his own definition of what was presented.
.
.
It's a reverse of the Transformers "robots in disguise" tagline, and quite appropriate given that Tesla named them "Optimus" (as in prime, the name of a transformer).
Even in the article:
"Another robot — or the human voicing it — told an attendee in a stilted impression of a synthetic voice, “Today, I am assisted by a human,” adding that it’s not fully autonomous. (The voice stumbled on the word “autonomous.”)"
So it seems Tesla were being fairly transparent about what was going on, and ironically it's this article chiding Tesla for being deceptive that is deceptive.
It comes across as a desperate reach to be critical: why say a "stilted impression" of a synthetic voice? As opposed to a mellifluous and natural impression of a synthetic voice? It gets it backwards: it's synthetic voices that are often stilted.
And then why note that it stumbled on "autonomous"? Is this supposed to be some gotcha that it's controlled by a bashful human after all? But it just told you that a human is controlling it... some trashy journalism imho
If Trump wins, all of the investigations will simply go away.
If Trump loses, then there are already a number of SEC/DOJ investigations targeting Musk over FSD claims and how X was purchased that would see him face jail time.
It makes it look like a classic dogecoin pump-and-dump scheme which backfired when it became clear how completely illegal it was. I haven't followed it in detail though so am curious about other interpretations
In any case, from the article,
"Another robot — or the human voicing it — told an attendee in a stilted impression of a synthetic voice, “Today, I am assisted by a human,” adding that it’s not fully autonomous."
So it seems in fact they were fairly transparent about what was going on. The article just has a deceptive headline.
What we'll see going forward will be similar to the vehicle fleet: the humanoid fleet will be manned by real humans, learning from them for years and gradually taking over various tasks as they become rote. Human controllers will over time be able to handle increasingly more bots simultaneously, until relegated to a purely supervisory role.
Just as with the latest Tesla vehicle update the driver doesn't even need to keep their hand on the wheel anymore, but merely watch the road, and robotaxis with neither steering wheel nor pedals are now feasible.
The tasks Tesla suggested included babysitting, and now that we know that the robots are built for remote control first, I would feel pretty uncomfortable with letting my child alone in the same house with a robot like that. Vacuum cleaners are bad enough.
https://www.vice.com/en/article/ecovacs-robot-vacuums-hacked...
>remote control first
If you mean it's the typical operating mode for the final product, I doubt we can conclude that just from an early public tech demo.
Factory robotics is mostly just conveyor belts and pusher whatevers. Not even something resembling a droid of any kind.
There are also potential ethical issues around teleoperation as a form of labor arbitrage. Should employers be able to outsource low-skill physical labor to actual humans abroad to circumvent wage and immigration laws? Is the benefit worth the economic dislocation to society? If you say yes, ask yourself why even bother reshoring manufacturing at higher cost and with trade barriers.
Teleoperation is argued to be an interim solution that trains autonomous systems the way driving trains FSD but as of right now, given what we know about just how much data is needed for current imitation learning approaches to work for fine manipulation tasks, it doesn’t seem scaleable. Maybe there will be a modeling breakthrough.
My bet is on cheaper non-humanoid form factors and data collection that doesn’t require a robot being operated for every data collector in the field.
Having a robot bake A cake and then wash the dishes. Not sensical to build a custom bot.
Laundry and washing dishes is not just “putting things in a spot”. Perhaps you don’t fold your clothes or separate your clothes or scrub dishes but many people do. Even loading the dish washer is fairly annoying when it’s a large crowd. Cooking is of course a huge one.
The question is really just how much would you pay for a 24/7 live in maid that can cook, clean, repair, etc. with a real human you’re paying them for availability. With a remote operated robot you’re paying for just the labor hours and none of the downtime.
It seems kind of fucked up but I do think the economics could work. People do have servants today. I tend to imagine such a service would be cheaper than a servant even at $20k up front cost. I do agree the bipedal legs seem unlikely be cost effective.
If it's a robot that needs to be able to work in any room, including navigating stairs, I'd think being bipedal is the simplest way to achieve it. Maybe make it tripedal to make walking/balancing easier.
Vs a real life human stranger in your house that could steal stuff and let accomplices into the house.
I already have a dish washing robot. And one that washes and dries my clothes. Can you imagine if you had a humanoid robot that was in your laundry room scrubbing your clothes manually in a tub and then taking them out and hanging them up on clothes lines? That is what you get with humanoid robots.
Would you rather have a tiny vacuum that parks itself in the corner and disappears, or some 6 foot tall thing with arms and legs pushing your Hoover around all day?
Clearing the table, scraping dirty plates, putting condiments back in the fridge, packing away uneaten food as left overs, rinsing the dishes, loading the dishwasher efficiently, then unloading the dishes and putting them all away in the arbitrary places the go, are all tasks to that cant be done by any robot that isn’t humanoid in nature in some way.
The amount of pen caps, dropped food, discarded clothes, school bags, shoes, partially assembled legos, couch cushions, books, and other random bulky items that end up on the floor of a house with young kids makes the idea of robotic floor cleaning being a solved problem laughable.
My assumption is that an Optimus home assistant will be an order of magnitude cheaper than manual labour, which means it will be accessible to people who can’t currently afford a cleaner/maid but whose lives would be improved by having help with the daily workload of life.
This brings up two thoughts, I wonder if the advent of robots will lead to more gender equality as women currently bear the a significantly higher percentage of the domestic work load.
Also, autonomous robots are going to make even harder to convince my kids to clean up after themselves :)
I don't understand what about that task needs it to be humanoid?
It obviously needs various abilities that humans have - being able to move around, being able to control multiple "limbs" to manipulate the clothes, etc. But why couldn't it look like R2-D2 rather than C-3PO? Why couldn't it be a flying drone that has 4 clothes-folding arms? Or... whatever non-humanoid design could be conceived that works best?
The only "need" for it being humanoid would be if the kids (or adults, or animals) found it more acceptable to be around.
The limitations of R2 are pretty obvious if you try to imagine it. there’s probably some optimizations that could be made but it’s a sensible start imo
But even your reasons for preferring a humanoid are all reasons why humans are better than current-technology non-humanoid robots, not unarguable facts that humans are the ultimate design. As a couple of examples:
> "You probably need legs to traverse areas."
Even if legs are definitely needed, the animal kingdom shows that human legs are far from the only choice. Why only two? Why not legs that are 90% of the height of the robot rather than human proportions? Why not legs with the equivalent of 20 knees rather than 1 knee, or legs that feature wheels that are sometimes used, or...
> "You can’t realistically fly while safely doing mundane tasks."
Not if you were to take any existing consumer drone and add robot arms to it, sure, but there's no scientific reason it can't be made with future improvements to technology. We already have drones that can automatically avoid bumping into things, and that can counter the effects of wind to stay in the same position, and we have drones that would be safe to walk into (ones with covered blades, where the only injury risk is it flying into you hard enough). There's no reason that a future version couldn't be just as stable hovering in the air while doing something as being on the ground - it just needs to extend its stabilisation algos so that it's not just countering the wind, but also pushing in whatever directions required to counteract forces caused by whatever its doing. And it also doesn't even have to be flying while doing the task, it can fly to the clothes, then unwind its leg (or legs) and stand there while doing its work, or...
Humanoid is obviously appealing for the simple fact that, if it's developed to the point that it has the same (or better) physical abilities as a human (including balance etc) then we know it can fit into anything humans do because we already do it, and because we've built a world around us for human-shaped people. But thinking there couldn't be alternative form factors that are just as good if not better is just lacking imagination on the subject of what technology will be able to do in the coming years - especially when not talking about a general purpose "can do anything a human can do" robot but about specific tasks (such as the clothes sorting & folding that we're discussing).
Obviously you can design better forms for specific tasks, but people don’t want some zany futuristic world populated by dozens of task specific robots.
But even for general purpose robots, my points above stand that future tech will mean plenty of non-humanoid shaped robots could be just as effective as general purpose robots as humanoid ones.
The ease of swapping different expertise in to the same body as needed could also potentially be useful.
> If you say yes, ask yourself why even bother reshoring manufacturing at higher cost and with trade barriers.
Mostly the reasoning to onshore things is to ensure independence on manufacturing critical resources despite the competitive disadvantage; not a jobs program.
So you'll have upper-middle class people own a robot and whenever they want a task done they order it from basically a call-centre somewhere in the world... reminds me of the etymology of "robot" - from the Slavic for "slave"...
You're right there are ethical issues, but hardly new ones for this context. In response to your question,
> Should employers be able to outsource low-skill physical labor to actual humans abroad to circumvent wage and immigration laws?
This is a very pertinent question ethically, but this already undergirds our current globalized society, where the the clothes most of us are wearing, the coffee we've been drinking, much of the food we've been eating today, and so on, are produced exactly by such labor. This is the way the world works, for the time being. And it's unfair. But these robotic possibilities don't add to the unfairness in my view, although they definitely would give it a new twist. Given current trends I wouldn't bet against increased twistedness..
Honda is now valued higher than Tesla, right? Oh wait... no they are not.
It was never remote controllable by unskilled operators.
It doesn't take a whole lot of imagination to come up with scenarios where even a remote controlled bot would be incredibly useful to society. If Honda never productized it or developed Asimo into something practical, then we can only speculate the reasons why. But that's on Honda for failing to follow through, ultimately.
I don’t think it really matters. The plan is to train the robots via telepresence. If that works out, no one will care that these demos used exactly that.
This statement is deliberately worded to avoid making a promise because Elon knows it is effectively a big fat lie. To date Tesla have solved the same problems that others have already solved. You can’t extrapolate progress because what comes next are the really hard problems that no one has solved. Even if Tesla is able solve those problems, there is zero chance that they can move at the same speed. They have 50% of an autonomous robot, but the next half is going to take 90% of the effort.
This is the usual smoke and mirrors. Elon shows off a tech demo using incremental gains and falsely represents how long it will take them to deliver on a revolutionary product.
His other companies less so. Tesla was early in EVs and did a lot of things right but the FSD push has been a huge disappointment and stuff like Cybertruck is a joke. Meanwhile other car companies are catching up. Boring company is boring. Maybe Neuralink?
The CyberTruck is the highest selling electric truck, so it's not a joke by any means.
Would you listen to yourself nitpicking at whether or not someone's 8th or 9th company is as impressive as their 3rd or 4th. Who has that kind of record, either present day or any time in history? Edison is about the only name that comes close.
So, everybody was making electric vehicles before Tesla? This is not how I remember it.
This was in response to California pushing for more efficient cars with the ultimate goal of zero-emissions.
What really helped Tesla was that they started out going for the sports car crowd. For someone's day to day workhorse car a lack of good support infrastructure is a big problem. People want that car to be useful for everything.
The people who could afford a $100k+ sports car could afford another car for when they needed to take the whole family somewhere, or take a long trip, or get groceries (the first generation Tesla Roadster's cargo capacity was less than that of a Mini Cooper convertible, about 1/3 of a Honda Civic or Nissan Sentra or Toyota Corolla). This meant that Tesla's market in the mid-2000s cared a lot less about EV infrastructure than the people who might have been interested in those '90s EVs.
There was another company making an EV sport car at the time, the AC Propulsion tzero. The people who founded Tesla actually founded Tesla because they saw that such a car would have a good chance of success but AC Propulsion wasn't interested in going into commercial production. They wanted to remain a technology company selling EV technology to others, rather than become a car company. So Tesla was formed, and licensed AC Propulsion's drive train and thus was born the Tesla Roadster.
You're better off with job specific designs like what Boston Dynamics does.
The fact that companies haven't been rushing to produce humanoid versions should tell you everything about their usefulness.
Where?
What makes it better enough to make it exciting when ASIMO stalled?
There's several videos (in the article, even!) of people at the event asking them if they are being operated by a human. They said that they are.
The Verge is trying to manufacture controversy.
One thing that the stock market investors don’t realize yet is that CEOs are not the most important part of the company.
The same “Shit of Theseus” phenomenon happens with companies like Blizzard which talent remains only in name and brand recognition, peharps would not be best to move on from expecting things from a brand name that constantly misses the mark.
Drop-in compatibility with all human-shaped legacy interfaces, for the most-rapid deprecation of homo sapiens as soon as the ASI can teleoperate humanoid proxies.
There's also the other argument about humanoid robots doing better in environments adapted to humans. But that's so far that hasn't really panned out.
This is the reason that every single car built have "two eyes" and "four legs". They could have one, or three or five. A car designed by insects would have six wheels, but then people would reject the model as ugly. All cars have an upfront and a posterior "face". And this face depicts either a mammal or a person. Never a bird or a snail.
As for headlights, a single central one would still be notably face-like given widely spaced turning indicators. And having two headlights gives redundancy: "one-eyed monsters" at night are preferable to cars lacking any lights.
I doubt this. There are physics issues involved with 6 wheels.
First off, where are those extra wheels going? Dually pickup trucks exist, but that's just so they can carry heavy loads.
If you put them anywhere other than close to the existing tires, unless you attach some sort of steering mechanism, or perhaps use casters, turning becomes a huge problem as those tires would have to skid.
No, we use 4 tires because it's the most stable and efficient number of tires to use unless you need to carry a ton of extra weight.