Assuming that models continue to get 90% more efficient every 18 months, and assuming that hardware continues to get 2x more powerful per watt every 2 years...
Aren't we still like a decade away from an Argon model being able to run on battery power for any reasonable amount of time that would fit on a robot that isn't the size of a school bus?
I'm naive, but wouldn't industrial applications be enabled by this anyway? You'll have in-house inference hardware that a robot can talk to via local network.
Also, your model probably doesn't need like a lot of its knowledge to control a robot.
I think for 95%+ of applications.. you just need a link to a data center anywhere.
My assumption is the hardware will have one/several smaller models running on it for locomotion, balance, etc.
Then the hardware "phones home" for decision making and planning.
Essentially the planning doesn't need to be close to the device as long as the local models can handle some part of the plan execution and make sure the robot doesn't fall over. The latency shouldn't be an issue. In fact more powerful centralized compute quickly creates an advantage over slower local compute in terms of end to end planning.
Liability issues will exist regardless of whether compute is on-device or central.
However, if the robot is in a contained environment like a home, there is less third party liability than for something like vehicles. Not none, of course, but less. And when you can get your arms around third party liability, first party liability can be better controlled by contractual agreements.
In general I expect on-device motor execution will allow it to safely come to a stop or enter a safe position if it loses the link to the execution planning cloud model.
> Aren't we still like a decade away from an Argon model being able to run on battery power for any reasonable amount of time that would fit on a robot that isn't the size of a school bus?
Sooner or later, the models will stop improving. After that happens, the best models will be etched into model-specific ASICs with weights fixed in mask ROM (like Taalas is already doing), which will result in massive, at least order of magnitude inference efficiency increase.
Transistors doubled for so long, but pretty much no other technology has come close, so it’s not like going to improve for very long, which is far more common. Transistors only did it due to a quirk in physics, that of electrons being long lasting, super tiny, and fairly controllable. We’re out of such particles to make another set of leaps.
For data bound things, there’s also fundamental laws of information theory, far stronger than physics theories, since they’re theorems about math. And LLMs can only compress knowledge so far, and we’re about at it now.
You can already run top of the line models on something that isn't the size of a school bus. The couple of KW you need to run them are the bigger problem, though on very high end hardware you might get double the efficiency.
Robots stand-alone AI do not really benefit from things like batch mode so you want to keep your LLM as high level as possible and your robot's subsystems as autonomous as possible, very light coupling between the two, definitely nothing time-critical. And that's the challenge: in robotics almost everything is time critical.
I'm a robotics noob currently building a prototype of a snow sweeping robot (www.frost-e.com) using my own design and I'm encouraged to see similar work coming from real places. Namely:
- Zero shot control using a visual harness and action primitives. I've seen this from a few other places as well (https://tml.stanford.edu/homebody/) and it's essentially having a generalist LLM command a robot via a set of predefined actions. Many robots and many tasks can probably be done with this approach, since not everything needs super intricate 6DoF arms or dexterity
- simulation bring-ups: I have a pipeline that takes in satellite imagery and reconstructs the lots and houses in a 3D environment and snows on it so that my virtual robot can learn to clean. With these latest models, you can probably create a higher fidelity digital twin more directly.
- self-improvement: These models (Argon, Astra, etc) can't operate the robot itself, but would be able to create data to distill into smaller models you can actually utilize.
Of course (the watermark is in the lower right), it's just a visualization. I'm spending my money buying parts for the prototype instead of professional renders.
You need to disclose that more clearly, at first glance I thought your actual prototype was pictured in the video until I noticed the snow just magically disappeared. I didn't even notice the watermark, and even if I had I wouldn't know that the random 4-pointed star meant "AI Generated Video".
I'd much rather see videos of a prototype than a fictional video that has no bearing on how the thing actually works. Especially since you list pricing on your website, making visitors think it's a currently-operating business.
As I approach a transitional period in my career, I'm reflecting on the managerial effort of places like google. How on earth do they keep an engineering org on the 'right track'? I work on products that are not as complex: fewer moving parts, fewer egos, no where near as novel. It feels like nobody does anything and nobody has any expertise
It’s like the USA. How do they keep people on track? They don’t. You delegate and hope things work. When they have Google Ads and Search they can experiment a lot and “waste” a lot in search.
They haven't solved management and definitely don't keep things "on track" globally. However, they definitely don't have the problem that nobody does anything and nobody has any expertise.
It's also worth noting, Gemini 3.5 Pro never came out.
Having worked there (but only in a small part of it that eventually got absorbed into Gemini), it's some combination of:
* The eye of Sauron (management) can make specific spots (e.g. Gemini) push harder (and attract the motivated people at the company and be selective)
* They hire a lot better than average (Generally from the top 10%, hotter places in Google are more selective)
* They have a promo process that gives career motivated people a way to prove themselves (not always aligned with things being on the right track ofc, and not always non-political, but they definitely do something)
* Redundancy in teams causes them to compete (not necessarily productively)
Of course, there are other downsides not listed here (insane bureaucracy for launches, GDPR/BCID compliance, etc, etc), but a lot of what makes Google possible is the insane amounts of cash Search throws off.
I am very new to robots and recently decided to go crazy and buy a humanoid from Unitree. I've spent a while building some tools in order to integrate LLMs with vision capabilities that can control the robot motion in a simple manner... and while I saw Opus and Astra crashing against tables, beds and surfaces (yeah, my LIDAR data isn't being used yet), Argon 4 takes a lot of precautions to avoid crashing against walls and other objects! I told my robot to go to a table, and the guy started walking slowly to the table, started steering at it and described the objects in it. Pretty scary for being something I've vibecoded in a matter of hours at home...
That sort of side project hobby could easily pay off in a nice salary over the next 10 years. It's like learning to make cars in the early days of automobiles.
There are a lot of guys with garages that have a $40k CNC mill in it. If you've got the money -- and in America, a LOT of people have the money -- $30k to be able to experiment with a fairly open platform that may play, as a category, a pivotable role in humanity for the next 15+ years is not very crazy.
New trucks optioned out cost $80k+ and plenty of people pay for that without having much justification of any kind for it.
> New trucks optioned out cost $80k+ and plenty of people pay for that without having much justification of any kind for it.
In the US, certain purchase categories make taking on debt to cover upfront costs trivially easy. Open robotics platforms are not one of those categories, while automobiles are.
It's a misallocation of resources. Actual roboticists can go their entire careers without touching hardware like that, and if they want it they have to spend years writing grants to get ONE to split between three other researchers and 20 grad students.
Meanwhile a guy who made a fortune in tech optimizing ad placements (not saying that's OP) can pick one up on a lark, and as you said it usually ends up in the garage next to other expensive toys like an $80k truck or autonomous Tesla. Also the $80k trucks are a great example because they're more of a status symbol, just like the robots will be.
Of course they used go their whole careers without seeing these robots, they've only been around for three years and used to cost hundreds of thousands of dollars.
These days, $30k is cheap enough for a university lab and so a lot of them have picked up these models.
It's 1.23 and weights ~30kg. Even though only the R1 Edu can be in theory programmed, it can be easily modded to enable SSH and root access to deploy custom movement policies and interact with the different on-board daemons to control the movements, grab the video data, etc.
I bought a second-hand Intel DeepSense camera and plugged it on one of the spare USB ports available in the main board to have 3D sensing (even though I really need to play a bit more with it). Right now I am playing with MuJoCo for RL training of new policies, integration with VLAs like NVIDIA Isaac GR00T and integration with LLMs.
I have to say its being a wild ride, and despite the small size of the robot, I have already injured myself with it when playing with the walking policies (If you activate a policy while its hanging, it starts literally kicking and punching you...) I cannot image those owning with the bigger brother of it, the G1...
Do you think the R1 you got can actually be taught to clean dishes and put them in a dishwasher in their specific sections? I see it doing so in https://www.youtube.com/watch?v=k1JIyXM5qGw
What's the latency like? Is the llm used to create an instruction set for the robot? How is it interrupted for reflex actions like controlling stumbling?
I'd love to play with this but something freaks me out: we heard no later than yesterday hackers ruining tens of millions of food by heating connected fridges in supermarket in France.
What happens when people decide to hack the Unitree robots and have them pick up knives and stab their owner and then go on a spree in the street?
Before people say: "But that's never going to happen, nobody would ever do that!" I'd like to state again the fact that people did just remotely hack fridges to ruin food.
"But ruining food for everybody is nothing, it's just kids playing around, people would never hack robots to hurt other humans!".
Is that what people really think?
Or do they think that "this time programmers managed to create secure software that cannot be hacked into"?
Hacking into what exactly? It's entirely up to the user if they want to connect their robot to an insecure webserver control hub.
The difference between you and most professionals is that we don't play. Groceries weren't playing with their refrigerators, they were attacked, the government can find and punish or kill the criminals if they continue their pattern of failure to bring justice the world will move past them.
It’s mildly impressive, but Google should be wiping the floor with its competition with all the resources it has. This intelligence isn’t even as good as figure 3. Unitree robots are basically doing parkour while this thing has to be hung up by a hook.
Google already has a dedicated robotics line called 'Gemini Robotics,' which is designed to support multiple robots. And itt's important to distinguish that the Argon implementation targets a completely different set of tasks compared to those robotics-specific models. Similarly, Unitree's parkour model isn't designed for manipulation tasks.
Is there something one could try out with not crazy costs?
E.g. for $250? I've got some rpi 4 (and rpi zero w) laying around waiting to be used if that helps.
Seeing prices like 5k EUR is a bit too much for a hobby.
54 comments
[ 2.1 ms ] story [ 10.4 ms ] threadAren't we still like a decade away from an Argon model being able to run on battery power for any reasonable amount of time that would fit on a robot that isn't the size of a school bus?
Also, your model probably doesn't need like a lot of its knowledge to control a robot.
My assumption is the hardware will have one/several smaller models running on it for locomotion, balance, etc.
Then the hardware "phones home" for decision making and planning.
Essentially the planning doesn't need to be close to the device as long as the local models can handle some part of the plan execution and make sure the robot doesn't fall over. The latency shouldn't be an issue. In fact more powerful centralized compute quickly creates an advantage over slower local compute in terms of end to end planning.
However, if the robot is in a contained environment like a home, there is less third party liability than for something like vehicles. Not none, of course, but less. And when you can get your arms around third party liability, first party liability can be better controlled by contractual agreements.
In general I expect on-device motor execution will allow it to safely come to a stop or enter a safe position if it loses the link to the execution planning cloud model.
Sooner or later, the models will stop improving. After that happens, the best models will be etched into model-specific ASICs with weights fixed in mask ROM (like Taalas is already doing), which will result in massive, at least order of magnitude inference efficiency increase.
For data bound things, there’s also fundamental laws of information theory, far stronger than physics theories, since they’re theorems about math. And LLMs can only compress knowledge so far, and we’re about at it now.
Robots stand-alone AI do not really benefit from things like batch mode so you want to keep your LLM as high level as possible and your robot's subsystems as autonomous as possible, very light coupling between the two, definitely nothing time-critical. And that's the challenge: in robotics almost everything is time critical.
- Zero shot control using a visual harness and action primitives. I've seen this from a few other places as well (https://tml.stanford.edu/homebody/) and it's essentially having a generalist LLM command a robot via a set of predefined actions. Many robots and many tasks can probably be done with this approach, since not everything needs super intricate 6DoF arms or dexterity
- simulation bring-ups: I have a pipeline that takes in satellite imagery and reconstructs the lots and houses in a 3D environment and snows on it so that my virtual robot can learn to clean. With these latest models, you can probably create a higher fidelity digital twin more directly.
- self-improvement: These models (Argon, Astra, etc) can't operate the robot itself, but would be able to create data to distill into smaller models you can actually utilize.
I'd much rather see videos of a prototype than a fictional video that has no bearing on how the thing actually works. Especially since you list pricing on your website, making visitors think it's a currently-operating business.
I think someone should make a game based on this concept.
is it just 'Hire the right people'?
It's also worth noting, Gemini 3.5 Pro never came out.
Having worked there (but only in a small part of it that eventually got absorbed into Gemini), it's some combination of:
* The eye of Sauron (management) can make specific spots (e.g. Gemini) push harder (and attract the motivated people at the company and be selective) * They hire a lot better than average (Generally from the top 10%, hotter places in Google are more selective) * They have a promo process that gives career motivated people a way to prove themselves (not always aligned with things being on the right track ofc, and not always non-political, but they definitely do something) * Redundancy in teams causes them to compete (not necessarily productively)
Of course, there are other downsides not listed here (insane bureaucracy for launches, GDPR/BCID compliance, etc, etc), but a lot of what makes Google possible is the insane amounts of cash Search throws off.
They don't.
If they did, neither OpenAI nor Anthropic would even exist, as the core of what they're all doing was invented by Google engineers and researchers.
New trucks optioned out cost $80k+ and plenty of people pay for that without having much justification of any kind for it.
In the US, certain purchase categories make taking on debt to cover upfront costs trivially easy. Open robotics platforms are not one of those categories, while automobiles are.
Meanwhile a guy who made a fortune in tech optimizing ad placements (not saying that's OP) can pick one up on a lark, and as you said it usually ends up in the garage next to other expensive toys like an $80k truck or autonomous Tesla. Also the $80k trucks are a great example because they're more of a status symbol, just like the robots will be.
These days, $30k is cheap enough for a university lab and so a lot of them have picked up these models.
It's 1.23 and weights ~30kg. Even though only the R1 Edu can be in theory programmed, it can be easily modded to enable SSH and root access to deploy custom movement policies and interact with the different on-board daemons to control the movements, grab the video data, etc.
I bought a second-hand Intel DeepSense camera and plugged it on one of the spare USB ports available in the main board to have 3D sensing (even though I really need to play a bit more with it). Right now I am playing with MuJoCo for RL training of new policies, integration with VLAs like NVIDIA Isaac GR00T and integration with LLMs.
I have to say its being a wild ride, and despite the small size of the robot, I have already injured myself with it when playing with the walking policies (If you activate a policy while its hanging, it starts literally kicking and punching you...) I cannot image those owning with the bigger brother of it, the G1...
What's the latency like? Is the llm used to create an instruction set for the robot? How is it interrupted for reflex actions like controlling stumbling?
What happens when people decide to hack the Unitree robots and have them pick up knives and stab their owner and then go on a spree in the street?
Before people say: "But that's never going to happen, nobody would ever do that!" I'd like to state again the fact that people did just remotely hack fridges to ruin food.
"But ruining food for everybody is nothing, it's just kids playing around, people would never hack robots to hurt other humans!".
Is that what people really think?
Or do they think that "this time programmers managed to create secure software that cannot be hacked into"?
The difference between you and most professionals is that we don't play. Groceries weren't playing with their refrigerators, they were attacked, the government can find and punish or kill the criminals if they continue their pattern of failure to bring justice the world will move past them.
Seeing prices like 5k EUR is a bit too much for a hobby.
Unfortunately the price when including shipping and taxes is fro me 455 EUR, which is twice my budged :(