If the point is to demonstrate the training (which is copying a human) then why hide the human off screen? Because it creates the impression that it's autonomous.
Putting "but it's not autonomous" in a subsequent tweet is just plausible deniability.
His fans will argue this is just a simple misunderstanding. C'mon guys. You used to be smarter than this.
The text of his tweet also "gives away the magic trick":
> Optimus cannot yet do this autonomously, but certainly will be able to do this fully autonomously and in an arbitrary environment (won’t require a fixed table with box that has only one shirt).
I just think it is entirely unimpressive. "Remote hands" robots, even with pretty fine-grained dexterity, are not at all uncommon in a huge variety of industries. There's zero autonomy here.
Fool me once, shame on you. I'm still waiting for 2015's "solved problem" of FSD (Musk: "I view it as a solved problem. We know exactly what we need to do and we will be there in a few years.") - though in "fairness" to Musk, apparently somewhere in the last few years, Tesla must have "lost" the solution. In 2022, "Our focus now is just on working on solving this problem".
What's your baseline? Just the fully custom actuation alone is impressive, to me, Especially so for the rates of progression, like, compared to Boston dynamics.
And, I don't think comparing the two is logical, for price (Da Vinci is $2,000,000), end goal (the 500kg Da Vinci will never walk across the room and pick up a box), or even degrees of freedom.
You must have missed the memo that the $35k Cybertruck is now the $60,990 Cybertruck (this is the standard-range, available in 2025 model). You will never be able to buy one new from Tesla for anywhere close to $35k, and the fact that they announced the price as $35k _and took customer money_ based on that should, IMHO, result in financial penalties and jail time for the executives involved.
I'm sorry, what does the CyberTruck have to do with the conversation, or their fairly obvious progression in actuation, or progression in making a nice humanoid platform for machine learning use, or Da Vinci?
It means the gap between what Musk promises will be available, and the cost thereof, versus the reality of what is available, when, and for how much, is "rather sizable" at times.
I'm still not sure how Da Vinci relates, or the CyberTruck.
But, sure, I don't really pay attention to timelines, since I don't have it on pre-order. They're making great progress. They will throw ML/AI at it, and it will be interesting to see what comes out. For people that enjoy tech, this is neat.
"They're making great progress. They will throw ML/AI at it, and it will be interesting to see what comes out. For people that enjoy tech, this is neat."
-Me, hearing Elon in 2015 describing fully autonomous self-driving as a "solved problem" and stating that "we will have complete autonomy in approximately two years."
> And, I don't think comparing the two is logical, for price (Da Vinci is $2,000,000), end goal (the 500kg Da Vinci will never walk across the room and pick up a box), or even degrees of freedom.
How much does a Tesla robot cost and where can I buy one?
Engineering doesn't ever involve starting with a final product. We're witnessing progression, nothing more nothing less. Progression can be observed and "appreciated", on its own. Working towards something "neat" is the modus operandi of most users here, and for some of us, seeing others working towards something "neat" is fun to watch.
I agree with this sentiment, and would likely 100% agree in this particular case if this were being done by some brand new startup. But at this point, Tesla's history is so littered with broken promises, truth-stretching, and outright lies that by default I treat any new announcement from them not only as marketing theater, but as a deliberate attempt to actively deceive.
A followup tweet with a fraction of the views with a pretty big delay. Was he going to add that initially and forgot, or did he add it because people started pointing out what was happening in the corner?
If this can be used for training then this magic trick can be enough. Training a language model uses in principle also just one magic trick - trying to predict a next sentence in a large set of ordered sentences.
> Optimus doesn’t appear have capabilities beyond anything we could do in 1964
Seems a bit disingenuous. Did the Lincoln robot have the dexterity in its fingers to do fine-motor tasks like folding laundry?
This has the same energy as saying the Falcon 9 isn't a massive innovation because the Saturn V existed in the 60s. Propulsive landing was done back in the 60s by the Apollo LEM, so obviously the propulsively landing re-usable first stage was "already done before".
https://mobile-aloha.github.io/
https://www.trossenrobotics.com/aloha.aspx
https://venturebeat.com/automation/stanfords-mobile-aloha-ro...