Humanoid robots coming soon, initially under remote control
- Reference: 1734593411
- News link: https://www.theregister.co.uk/2024/12/19/humanoid_robots_remote_contral/
- Source link:
The first message over the internet in 1969, then known as ARPANET, was "LO" – which would have been "LOGIN" had the system not crashed.
And the first robotic foundation model API call in 2023 was, "Put the eggplant in the pot," according to Sergey Levine, co-founder of Physical Intelligence.
[1]
Levine kicked off the [2]Humanoids Summit at the Computer History Museum in California, an event intended to explore the state of humanoid robotics and artificial intelligence.
[3]
[4]
As far as milestones go, vegetable placement lacks drama, but it represents a meaningful marker on the road to making humanoid robots relevant to the general public. Manipulating aubergine without incident is not a trivial technical challenge.
Roving robots have already arrived for those in San Francisco, California; Phoenix, Arizona; Los Angeles, California; or Austin, Texas, in the form of Waymo autonomous vehicles.
[5]
Humanoid robots have made a showing in laboratory and factory settings. Fourier [6]debuted its GR-2 line of humanoid robots in September. Figure's O2 humanoid robot has been [7]deployed by BMW . And Boston Dynamics' Atlas has starred in various videos doing stunts.
[8]
Image of a Fourier GR1 robot – Click to enlarge
Soon - the exact date to be determined - human-like bots will be available to the general public.
Based on slides presented by Bernt Børnich, cofounder and CEO of 1X Technologies, the firm's [9]Neo humanoid robot is scheduled to debut in 2025 for a set of early adopters in the San Francisco Bay Area.
That comes with a caveat, however. Neo, in its initial rollout, will not be fully autonomous. Rather it will be teleoperated.
"It will have some tasks that will be done by AI," explained Jorge Milburn, VP of sales for 1X. "Navigation, for sure, pick-and-place, and other stuff will come out of the box."
[10]
But, he said, Neo will be paired with a human teleoperator capable of directing the robot from afar – probably on a one-to-one basis. "Over time, the idea is more supervision than actual tele-op," he said.
Milburn said 1X is likely to be conservative about where it places Neo initially – meaning homes with children would be avoided due to the higher safety concerns. Neo is only 5' 4" and 66 lbs, with four hours of battery life, but it could still do damage if it fell on a child. The biz is looking specifically for customers who want to provide feedback and help it gather the data to improve Neo.
The main bottleneck is the diversity of data ... The path to widespread adoption in society for robotics is going to be through the home, because you need that data
Beyond assisting with various mundane tasks, Milburn said he expects the companionship aspect of Neo will be important, which would entail conversing either with the resident AI model or the human teleoperator.
"We will also probably offer the ability to teleport yourself," he said, meaning that those away from home with a VR headset could see through the eyes of their home-bound bot and presumably take over some teleoperation or communicate.
The limited autonomy of the first robots headed into people's homes follows from the fact that, like autonomous cars, a lot of data is required for a machine to operate in a residential environment – and that data has not yet been gathered. Simulated data can help, but for humanoid robots to navigate effectively in the human world, they have to stumble and fumble before they can walk among us.
"The main bottleneck is the diversity of data," said Børnich. "The path to widespread adoption in society for robotics is going to be through the home, because you need that data."
[11]Letting chatbots run robots ends as badly as you'd expect
[12]Robots crush career opportunities for low-skilled workers
[13]Oh, what a feeling: Toyota building robots that get better with practice
[14]US Army orders next-gen robot mule to haul a literal ton of gear
Just as Waymo robotaxis initially were accompanied by a human supervisor, Neo bots and other domestic human-like machines will head out into the world, awkwardly, under human oversight, to gather the data required to allow them, eventually, to operate on their own. And as with mobile phones, expect that the privacy policies imposed by robot makers will not afford much privacy – the robot's video record of dropping the breakfast you requested will help company engineers improve its performance.
Levine's firm, Physical Intelligence, has been working on an AI model that should help robot makers move beyond teleoperation training wheels. It's a general-purpose robot foundation model called π0 (pi-zero).
It's not just a vision-language (multimodal) model along the lines of Google's Gemini 2.0 Flash. It's a vision-language-action model, meaning its responses can be translated into commands that control robot hardware – similar to a software agent system that has access to physical hardware rather than software applications.
"The concept for a robotic foundation model is a model that is trained on data from a huge range of different robotic systems performing a wide variety of different tasks," Levine explained.
"It's pre-trained on internet-scale data that gives an understanding of the vision and language. And it can then be prompted to perform tasks or fine-tuned to particular application domains with a lot less data or a lot less effort to take a spin off those applications from scratch."
This approach has been shown to work – a robot armed with the π0 model can [15]fold laundry, bus tables, and assemble boxes . Human workers are still better and faster at these tasks, but robots are improving.
The employment consequences of humanoid robots came up in various panel discussions. Much was made about aging populations and jobs going unfilled as a rationale for developing capable robots. Those selling mechanized labor often said their machines would take jobs no human wanted to do or would simply help people focus on more fulfilling aspects of their jobs.
A lot of the humanoids we see out there are a hammer in search of a nail
No doubt there will be some of that, but it's equally true that companies have shown little loyalty to human employees when they have the opportunity to reduce labor costs through offshoring. Where robots can do jobs more affordably than people, with at least the same level of competence and efficiency, expect humans to lose out over time.
Some skepticism was raised about whether the humanoid form factor is the right choice, given that in some scenarios, a fixed robot arm or wheeled machine might make more sense.
"A lot of the humanoids we see out there are a hammer in search of a nail," observed Leo Chen, director of US operations for Engineered Arts, during a panel discussion.
Chen pointed to the Boston Dynamics Atlas robot, one of the most advanced robots out there, saying that the most revenue the company earned from it came from entertainment – from putting it in [16]a Super Bowl ad – rather than from its ability to perform any practical application.
It would be really, really, really nice to get some diversity in our robot designs
As far as the tech industry is concerned, the right form factor is whatever customers will buy. And certainly for scenarios that require operating in environments designed for people, there's a good argument for making robots in our image.
The form factor of those making robots also came up. Ross Mead, founder and CEO of Semio, observed that most humanoid robots reflect his boyhood ideas about what robots should look like. Just as the six people on his panel were all men, he said, current robots tend to reflect the design priorities of dudes. "It would be really, really, really nice to get some diversity in our robot designs," he said. ®
Get our [17]Tech Resources
[1] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2Z2P83jfmiQq7f-id6OBKVQAAAQM&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[2] https://humanoidssummit.com/
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44Z2P83jfmiQq7f-id6OBKVQAAAQM&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33Z2P83jfmiQq7f-id6OBKVQAAAQM&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44Z2P83jfmiQq7f-id6OBKVQAAAQM&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[6] https://www.fftai.com/newsroom-newintech/14
[7] https://www.press.bmwgroup.com/global/article/detail/T0444265EN/successful-test-of-humanoid-robots-at-bmw-group-plant-spartanburg?language=en
[8] https://regmedia.co.uk/2024/12/12/g1_robot.jpg
[9] https://www.1x.tech/neo
[10] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33Z2P83jfmiQq7f-id6OBKVQAAAQM&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[11] https://www.theregister.com/2024/11/16/chatbots_run_robots/
[12] https://www.theregister.com/2024/11/08/robots_crush_career_low_skill/
[13] https://www.theregister.com/2024/10/17/boston_dynamics_lbm/
[14] https://www.theregister.com/2024/09/28/us_army_robot_equipment/
[15] https://www.physicalintelligence.company/blog/pi0
[16] https://www.youtube.com/watch?v=1B-PZTqJGQo
[17] https://whitepapers.theregister.com/
"Put the eggplant in the pot"
Not tonight darling, I have a headache.
Remote human operator!! Okay, it's not field ready, but as I say with cars: don't live test this shite on the public! FFS, you'd allow a complete stranger in to your home (albeit by proxy). What could go wrong indeed.
I'd rather wait until it's fully autonomous thanks.
I do want this tech to be pursued - my "True AI" approach absolutely requires the AI to have physical form (its own 'corpus', as opposed to a 45U rack!). YMMV.
Re: "Put the eggplant in the pot"
If you want the tech to be pursued, it needs to be trained and tested out in the real world with real people.
I'd happily have one in my home for testing, within certain boundaries - remote kill-switch, and the absolute ability to limit it to certain rooms or floors of the house, as a basic minimum. I'm sure I could think of others given time.
GJC
Re: "Put the eggplant in the pot"
Indeed. But not initially tested at consumer level (ie. in public) - business level, where insurances and lawyers can come into play.
[Edit: military bases, where they can be "retired" at will! :D ]
Re: "Put the eggplant in the pot"
IT was in niche academia/military/business for decades before hitting the mainstream of those areas. PCs were in businesses for (call it) 20 years before they became prevalent in the home. The wrinkles were ironed out to a great degree by then (honest!). Those preceding live tests were undertaken by [orgs] with the capacity to support/fix/develop/mitigate any glitches. They paved the way for domestic/public utility.
And none of them had an unknown/unvetted third party "on the inside" using a live link. A link that we accept can never be deemed 100% secure.
Granted, you and I, and a good number here, are the geeks and nerds that would happily pilot such devices - it's in our nature. But we (believe we) have the capacity to support/fix/develope/mitigate any glitches.
Re: "Put the eggplant in the pot"
All you are referring to there is the results of affordability, not whether the tech had been "ironed out" well enough to be deployed in greater numbers, either in business/military or in domestic settings. They were always being targeted at "The Home of The Future", with your Telex in the corner of the living room and a chair that allowed the dozens of underskirts to fit.
> PCs were in businesses for (call it) 20 years before they became prevalent in the home
That wasn't anything to do with (general use within) business ironing out anything - I'd even argue the opposite, computers used for playing games at home, or before that just in arcades, giving the wider early exposure to many more people than using them in their work lives (purely mechanical tills stayed around a long time).
Mass home computer adoption stems from availability of networking, the Web, monetising the games etc etc. These were pushed by, exploited by, specific businesses (exploiting the work of non-business types, as per usual) - well, ok, because many businesses could afford networking before there was any *point* in domestic installations, they did have their staff testing out online games, downloading porn etc before it hit the suburban home.
Re: "Put the eggplant in the pot"
Regardless of the motivation behind "domestic insertion" the period of extended testing took place.
Re: "Put the eggplant in the pot"
> period of extended testing took place.
Random usage in a different arena with a different set of use cases is not "testing".
Re: "Put the eggplant in the pot"
LOL - okay, the word of the law vs the spirit. I'll modify my statement.
Regardless of the motivation behind "domestic insertion" the extended period happened.
Whether or not it was testing to your satisfaction, data was gleaned and fed back beneficially improving the products.
Re: "Put the eggplant in the pot"
"PCs were in businesses for (call it) 20 years before they became prevalent in the homeW
Home computers became fairly common in the mid 80s. 20 years would imply business PCs were common in the 60s. I am pretty sure they were not. Mainframe terminals were pretty common, I believe, but that is an entirely different technology.
Re: "Put the eggplant in the pot"
I did specify PCs. I considered mentioning home computers and consoles, but omitted them as "limited functionality devices". Even so, if you want to start from there, it just shifts the timeline, not the point.
[Edit: Colossus 1944 -> Home Computers for Nerds 1975 (for argument's sake) -> some businesses mid-80s -> "all" businesses mid-90s -> A PC In Every Home 2000 (again...)]
Re: "Put the eggplant in the pot"
> I'd rather wait until it's fully autonomous thanks.
>> "It will have some tasks that will be done by AI," explained Jorge Milburn, VP of sales for 1X. "Navigation, for sure, pick-and-place, and other stuff will come out of the box."
So it already *can* be fully autonomous - it'll just be rather limited in what it can do. But if that was all your Use Case called for, job done. The same as for many, if not all, systems, robotic or otherwise.
Now, being able to perform absolutely *any* task (within its physical limitations) without external guidance or control (such as being teleoperated with its "learn this" switched on)? That is going beyond even human abilities and into the god-like: personally, I'm glad to have received training, even down to the level of "here, let me place your hands, now I'll move your hips and push your foot over there - good, now eye on the ball and try a swing".
Unless you are happy having your mechanical pal who is fun to be with being limited to the role as an overpriced roomba, coping with the messiest human environment, a home with kinder, teens, dotty old aunts and other agents of entropy, is going to require supervision of the android.
Now, whether you do all that extra supervision yourself (whilst the big birthday for the five year old is still going on) or hand the duties over to a paid professional (or to an external Corporate Master AI - does that make you feel safer?) - hmm, choices, choices. You've paid for the damn thing, gonna let it just stand there and rust?
Re: "Put the eggplant in the pot"
Offspring learn largely by mimicry - not all, some species never see their parents. Either way those species have undergone millions of years of evolution, their method of growth (early learning) has successfully survived the horrors of nature. There's no "ultimate controller" beyond their own species (remote operator or Corporate Master AI). Even if you want to argue God etc, I don't think said deity prevents you sticking your fingers into that wall socket!
Animals have tactile feedback. If you move that foot in the wrong way, or to the wrong place, it hurts! A pretty strong signal "not to do that again". Parents don't want to see their kids hurt, but they know they've got to fall to learn - one day they'll leave the nest, and repeat that process individually, hopefully well armed to deal with their next trip-up. We don't put our toddlers in their own apartment, we wait until they're a bit better equipped.
Parents don't remote control. (Well, some try, but then it takes all sorts.)
Re: "Put the eggplant in the pot"
> ...species have undergone millions of years of evolution...
Fascinating diversion away from the subject - expensive mechanical devices being used in the home.
Unless, unless - your willingness to wait until humanoid robots are ready to be deployed in the domestic setting means you are going to wait for millions of years for the androids to naturally evolve?
Not sure how the VCs will react to that long a wait for ROI.
Re: "Put the eggplant in the pot"
I wasn't diverting anything. You raised the point "I'm glad to have received training, even down to the level of "here, let me place your hands, now I'll move your hips and push your foot over there - good, now eye on the ball and try a swing".
And a valid point it is.
Our ability to run simulations can mimic those millions of years of evolution in a shorter time(?). But if that's not being done, the comparison to human/machine can never be drawn. If that comparison can not be drawn, our understanding of intelligence cannot be applied. (AI be the subject here.)
[Edit: or at least I thought you were referring to human learning - if not, apols. But the comparison stands.]
Generative feedback
This could be the turning point. To date, AIs have had little or no control over what happens after they spit something out. Robots which learn basic motor skills like a baby have been fundamentally unintelligent. A generative AI controlling a robot will spit out purposeful movements, and suddenly gets the chance to learn from its mistakes in a far more meaningful way.
However, the next couple of years will likely reveal yet another layer of sophistication to be missing - but what?
Re: Generative feedback
...another layer of sophistication to be missing - but what?
Sentience.
I really can't get over
These things will not be used for "good." They are built to invade, record, and collect. Soon, hackers will use them maliciously. I cannot believe people who claim to be so superior in intelligence can't see what's coming.
Or maybe.... you ARE the evil.
"The path to widespread adoption in society for robotics is going to be through the home, because you need that data."
At last - honesty.
"A lot of the humanoids we see out there are a hammer in search of a nail,"
Could be said for AI in general.
I also find the trend to humanise machines a bit tragic. Why do we want the tech we use to display human characteristics? Why can't we treat them as what they are, "machines"?
I know there is a tendency for us to form emotional attachments to the non-living things around us (cars are a prime example) but we should resist it. Only the marketing people trying to flog this stuff to us think this is a good idea.
Neo will be paired with a human teleoperator
So, how much is this going to cost? The teleoperator will have to be trained to operate the machine competently, which if it is to be in any way useful, will be quite complex. How many hours a day will the robot operate for? 24 hour operation would require 3 shifts of full time workers. They will require breaks - the robot won't be able to make your lunch if the operator is on their lunch break. Frankly, it would be more useful to send the operator round to your house to do stuff for you rather than trying to control the robot, as they would be able to do far more than moving aubergines small distances. As for having to lug a VR headset around with you everywhere, so you can "look" through it's cameras - who's going to do that?
Finally, it's LLM powered. The chances of it mishearing or misinterpreting an order are decidedly non-zero, and you wouldn't want to come home to find that it's turned your pet cat into a fluffy hat.
Perfect for amazon's RTO mandate.... Make the telepresence robots good enough, and the managers can go work in the office, the warehouse staff can work from home....