At 17:25 they state that droids have a 10 percent chance of doing a task they haven’t been trained on. They also state in the video that they can easily train a droid to do almost any task.
(First seen on Saul’s board)
Andy
2 Likes
10%
Does that mean if the task is made up of 2 actions it will screw up endlessly? He is only talking about context. Then reasoning is a zero for now. No kidding.
So spend more money. LOL
That is with no training Leap, that is why I say it is huge, if they train the robot it is almost flawless. It’s like having you do a task you have never seen or done, the chances you get it right the first time? The robot is training itself. Go to 38:20 and listen to that.
Andy
2 Likes
Did you notice they never named the company building this robot?
NPR discussion last night wound up saying “imagine…”
I think adding to your imagination is profitable.
Last year it was perfectly planned out cities where no one drove themselves. Where very little travel was done because everything was done at home and delivered automatically to the home. Last year’s imagination.
Leap the whole video is built on the company that is building the robot. It’s the founder they are interviewing. Its called 1X. I understand the video is long but we really can’t discuss it if you are not viewing at least the portions I gave you.
Andy
4 Likes
Fair enough but I was bored. You’ll have to forgive me.
Did you notice the smug faces on both of them? They were pulling one over on you. That is why I tuned out.
I do not buy into liars.
Andy, could you point me to the point in the video where they talk about the robots that are already being sold? I didn’t see that discussion in the portions you called out.
1 Like
Albaby they actually have a transcript of the video if you would rather read that. You can go onto youtube and down at the blue portion hit more, go down to the bottom and hit the button that says show transcript. One portion where they talk about commercial robots being sold you will see at time stamp 2:21 but if you go back to time stamp 2:02 you will see where they add context to the conversation.
Andy
1 Like
Al,
That was missing for me as well.
BTW robots being sold does not equal robots doing a single thing at all.
The success rate was 10%?
Thanks. The transcript is helpful.
Impossible to tell from the discussion whether there’s actually much “there” there. It seems like the 1x guy is talking about their EVE robot - which gets around on wheels and has teleoperability:
So…is a security company using a robot like that necessarily doing anything significant? The 1x CEO doesn’t say whether these things are operating autonomously or using the humanoid part of their form for anything - and if they’re being used by a security company, they wouldn’t need to be anything more than a camera on wheels. Which is useful, don’t get me wrong - but also makes the “humanoid” part of the EVE utterly cosmetic.
Try this YouTube?
The discussion of 1x, Eve and Neo starts just after 23:48 minutes. Eve is ‘wheeled’, while Neo has bipedal legs.
Also, note these guys are saying Tesla, Nvidia and Figure are all saying a robot will cost $10-20K.
At 20:58, they say that Fourier (Singapore, Chinese) company has robots for sale today.
ralph
1 Like
Again, same issue. EVE is wheeled, doesn’t have fingers (or even really a hand), and is teleoperable. So there’s no way of telling whether any of what it’s doing for these security companies is anything more than “camera on wheels.”
As for manufacturing costs, I have no doubt that these folks can manufacture a “robot” for less than 10-20K, because the body of a 100 lb or so machine isn’t going to cost that much to make. So if you don’t define what the brain and hands of the “robot” are, or what it’s capabilities are, I’m sure you can quite easily make one that costs less than a car. The real issue will be what it would cost to make a robot that’s got the brains and hands that would allow it to make use of its humanoid form - to do a task that can’t be performed by a camera on a wheeled tripod (for example).
1 Like
No. The success rate was on being able to do a chore it had not been trained on.
2 Likes
I don’t understand, why is it the same issue with NEO?
They have already sold 150 humanoid robots, How many wheeled tripod camera robots have they sold? Also as Ralph pointed out two companies, one from China the other from Singapore, are actually selling Humanoid robots on the internet. I think we are at an inflection point where it is only going to get faster and faster with adoption. Maybe I am wrong but there are so many companies making Humanoid robots now Its really hard to believe that they are not going to get them up to scale.
Andy
1 Like
Great video Ralph, thanks, I went to the beginning and listened to it and it was very succinct.
Andy
Maybe I missed it - did they say they had sold a bunch of NEO’s yet? In the discussion about sales where they were being paid for services provided (ie. not just pilots), they were talking about EVE’s.
They just did a fundraising round to get the equity to get NEO to market in January, so I don’t know that they’re actually selling any yet:
They’re not in the business of selling wheeled tripod camera robots - they’re in the business of selling “humanoid” robots. The question I always have is whether these “humanoid” robots are doing any jobs where it’s necessary (or even advantageous) for them to have a humanoid form factor, rather than that being an aesthetic choice. You can have a “humanoid” robot that has arms, and bends at the knees - but if all they’re being used for is to roll around as cameras, then those are just cosmetic features that help market the device but don’t really represent a practical application of the humanoid form.
And since these were sold to a security company - and not to (say) Amazon - it certainly seems plausible that they’re being used for monitoring rather than any other function. They’re certainly not going to be apprehending someone breaking into a facility.
Most security guards these days don’t apprehend people, the insurance liability is just to high. I think you need to listen to both video’s there are so many things you are missing.
Andy
It’s two hours of content - perhaps you could highlight a few points you think I’m missing?
Reading the transcripts, they don’t really show that humanoid robots yet have the capabilities for there to be practical applications for the humanoid form. The EVE robots don’t appear to be doing anything that actually requires any of the humanoid aspects of their design - as we’ve discussed, they’re basically just operating as remote cameras, moving around on wheeled platforms. I’m not even sure they’re doing that autonomously, since 1X CEO talks about them being teleoperated. So while it’s nice that they’ve found a place where the things they’ve built can provide some use, that’s not a job that needs to be done by - or even benefits from - the humanoid form factor.
Herbert Ong is very enthusiastic about this technology, but he doesn’t ask especially penetrating questions in these videos. These guys are talking their book and excited about their interests in robotics. But I get the same general vibe from them as reading supporters of blockchain (or even the Segway, back in the day) - they’re so convinced that this is the Next Big Thing and that it’s Right Around the Corner that they’re pulling together all the littlest signs of progress and treating them as revelatory.
Again, unless I’m missing something significant that might be buried in one of these videos that I overlooked while reading the transcripts. Anything?
4 Likes
Don’t worry about it, it will be here before you know it. Then you can get a close up view of it.
Andy
I certainly hope so. I guess it’s possible, but sometimes the Next Big Thing that’s going to leap from amazing technology into practical application…just doesn’t (or at least takes a lot longer than people think). Which is why my life isn’t filled with blockchain applications that I regularly access through augmented reality while riding in a self-driving car. Sometimes the tech is ready but there really isn’t a valuable use case that makes it worthwhile, and sometimes the use case is obvious but the tech takes far longer than early developers and enthusiasts imagine it will.
4 Likes