Hacker News Reader

Comment by bobxmax | Hacker News Reader

We are close. Language models and large vision models have transformed robotics. It just takes some time to get hardware up and running.

View on HN

I think it would be many decades before I'd trust a robot like that around small children or pets. Robots with that kind of movement capability, as well as the ability it pick up and move things around, will be heavy enough that a small mistake could easily kill a small child or pet.

That's a solved problem for small devices. And we effectively have "robots" like that all over the place. Sliding doors in shops/trains/elevators have been around for ages and they include sensors for resistance. Unless there's 1. extreme cost cutting, or 2. bug in the hardware, devices like that wouldn't kill children these days.

> Sliding doors in shops/trains/elevators have been around for ages and they include sensors for resistance

Some of these are pretty crazy too.

Here's a video from 14 years ago where a table saw stops fast enough that it didn't scratch a hotdog: https://www.youtube.com/watch?v=fq3o0VGUh50

So even if this hypothetical robot had saws for hands it could be mostly safe (in theory).

Even for adults, a robot that would likely have to be close to as massive as a human being, in order to do laundry and the like, would spook me out, moving freely through my place.

You'd learn to trust it. People have pet dogs that could kill them if they wanted. As can other humans walking around your house.

People also have essentially wild beasts in their home: cats. If cats were the size of small dogs they would kill people all the time, but we love them when they are small enough so they just claw you bloody.

Since we can live with that we can live with anything that doesn't outright murder us.

> have transformed robotics

Did they? Where? Seriously, I genuinely want to know who is employing these techniques.

All frontier labs are now employing LVMs or LLMs. But that's my point is you won't see the fruits of it this early.

That's the point being made. It's transformed robotics research, yes, but it both remains to see whether it will have a truly transformative effect on the field as experienced by people outside academia (I think this is quite probable) and more pointedly when.

I think this is an opinion borne out of weariness with constant promises that amazing robots are right around the corner (as they have been for 20 odd years now). For anyone who is close to the front line, I think the resounding consensus is clear - this time is different, unbelievably different, and capability development is going to accelerate dramatically.

I think it's impossible to spend a lot of time with these models without believing robotics is fundamentally about to transform. Even the most sophisticated versions of robotic logic pre-LLM/VLM feel utterly trivial compared to what even rudimentary applications of these large models can accomplish.

> have transformed robotics

When questioned:

> believing robotics is fundamentally about to transform

These are not even remotely the same thing. Something that has happened already and is verifiable fact is not the same thing as your opinion, even if your opinion is based on a lot of sound arguments and reasoning.

Very tiresome to read so many claims of fact based on opinion of what will happen in the future.

The discussion was about whether robotics was about to transform or not. And obviously it is because of how much basic robotics workloads improve with these models.

Really not that hard.

> have transformed robotics

> about to transform

Apparently even english tenses are too hard, let alone anything else. Bald faced lie, to claim what you think might happen in the future has already happened in the past. No matter "what the discussion was about", or what arguments you bring to support your estimation of the future.

https://www.figure.ai/

specifically their speech demo video (which is, of course, a demo video)

https://youtu.be/Sq1QZB5baNw

https://www.1x.tech/neo and

https://www.unitree.com/h1/

are undoubtedly using such models.

It's an area of active research, eg

https://www.physicalintelligence.company/blog/pi0

https://wholebody-b1.github.io/

https://ok-robot.github.io/

https://mobile-aloha.github.io/

Reply on news.ycombinator.com