58 comments

[ 2.6 ms ] story [ 109 ms ] thread
I wish this was a good critique of Optimus. But this is repeating a couple of witty quips from random people on Twitter. Is this journalism?
It was written by media experts so it must be.
(comment deleted)
I really don't get all the critique of the Tesla bot. Yes, it isn't even close to a finished product. But it isn't meant to be. Not yet. But it is the development platform for an upcoming product. And like with software production, there is nothing as valuable as a working prototype which you can use to test all the aspects of the ongoing development. They also showed the second generation robot, with a lot of improvements. Most of all, in house designed actuators which are fit for industrial production. Because that is the goal for Tesla: to eventually sell it. In numbers, at a reasonable price. Part of the purpose of the event was to attract new developers. So any of the Ph.Ds who think they can do better: apply at Tesla!

Also the comments about the finger count were revealing. They don't grasp the ambition of Tesla. They are not trying to quickly get a new warehouse robot to the market, their end-game is the fully humanoid robot. That means 5 fingers. And a real AI for guiding the actions. Like with FSD, Tesla is betting big time that they can get reasonably close.

The goal is to tingle the next shiny thing in front of investors to somehow keep the stock evaluation.
It's such a silly complaint. They've only had a working model for 6 months. Boston dynamics been at it 30 years almost.

Also Tesla bot doesn't ever need to do backflips so they are closer than ppl realize already

> They've only had a working model for 6 months.

Why should I take that into account? This isn't school.

Velocity is relevant.
Let's discuss it then, then. Velocity at outliers has little explanatory value.
What does school have to do with anything? In a short time they have made tremendous progress which will obviously only continue.
The Tesla bot was wheeled out on stage by three people, because the team wasn't sure if it could walk on its own.

Meanwhile, Boston Dynamics bot is doing backflips, and Disney's superhero bot is doing stunts for it's movies / theme parks live.

Tesla bot can't even walk, but there are two robotics groups (Disney Theme parks and Boston Dynamics) who were ahead years ago.

--------

If you wanna impress people who have been looking at bots for the last 10 years, you have to at least have the thing walk out on stage on its own feet.

No, it wasn't. That was the second gen bot which isn't fully operational yet. The first gen model walked onto the stage without support.
So you clearly didn't watch it at all because that's exactly what it did.
But Boston dynamics didn't show it to the public until a few years ago when it was good. It's not like we were all seeing Boston dynamics videos of their robots sucking 30 years ago, as Musk is doing.
Not true. Someone on reddit made a compilation of BD videos, with the first dating about 30 years.

I'm on mobile, try searching "30 years Boston Dynamics" on reddit.

Where did they release these videos at the time? YouTube didn't exist, the tech to distribute video on the internet didn't exist at the time period. I suspect that these videos you refer to were made after the fact from "home videos".
That's just not true they have been showing videos at all steps along the way, from WAY before they were at the level that the Tesla bot is now
Showing videos /= what Tesla did with this press release event.
So now your are complaining they did a demo? That is just silly. "It's wrong to do a demo" basically lol
I'm not saying it was wrong, he can do whatever he wants. I am saying that it shows he is not longer a legit tech leader, he's a seller of hype.
But Boston Dynamics didn't need to pump their stock before their owner cashes out....
> So any of the Ph.Ds who think they can do better: apply at Tesla!

Why not Boston Dynamics, who are clearly much further in this process and have much more expertise?

There's Boston Dynamics "Spot" that's clearing mines in Ukraine right now. How long will it be before this Tesla-bot can do what Spot does today?

Why would a Ph.D join a group with only a few months of experience, when a true expert is making revolutionary robots exists elsewhere? Ph.D want to learn from others, not be the smartest person in the room.

-------

Lets look at a 2018-era (4 years ago) demo from Spot: https://www.youtube.com/watch?v=wXxrmussq4E

Spot, years ago, could open a door by itself. How long before the Tesla-bot can do that? Why would any Ph.D want to be working on an already solved problem (solved years ago by Boston Dynamics), when they could just join Boston Dynamics and learn from the experts directly?

Or Honda, they already made one of these, and did it well. Then they canned the project because there is no commercial value. There is no need for a humanoid robot, a regular wheeled one will do and is 10% of the cost.
Boston Dynamics clearly has talent but it doesn't seem they've found that market fit yet. Softbank and Google both sold Boston Dynamics. It's not some scrappy startup, it's on its third multinational acquisition by now.
This is kind of an absurd premise? There are not that many companies working on this kind of Robotics, are you seriously suggest all US talent in the field should go to Boston Dynamics?

Boston Dynamics simply has a different focus. They focus is not mass production of humanoid robots for everyday use. Their primary concern is not using their mass manufacture and self driving expertise to mass produce robots.

Tesla is expert in electronics and their robot is driven by electric actuators while Boston Dynamics seems to use a lot of hydraulics. Hydraulics are like better in some ways but also likely harder to mass produce.

Multiple companies working on these topics is a good thing in my opinion.

> Ph.D want to learn from others, not be the smartest person in the room.

If Tesla has the funding to heir people, do research and build many robots then what's the issue? You don't get people with experience and you don't get teams with experience unless you start building things.

I'm sure there are people who would like an opportunity to be in a leading position in such a project. Even when in theory there are other projects that are 'more advanced'.

https://www.youtube.com/watch?v=QfbdVboVNUM

A better automatic appeltizer will change the world in a more meaningful way than a humanoid robot.

Or a chip-shooter.

https://www.youtube.com/watch?v=SRu02F6AOmg

If you really want a humanoid robot, Boston Dynamics is finding applications for walking robots, and Disney is finding applications for superhero robots for their Theme Parks / Movies.

https://www.youtube.com/watch?v=ENa98h7M7qY

-------

The rest of the world is very, very, very far ahead of Tesla here. What Tesla demonstrated is not very special.

> A better automatic appeltizer will change the world in a more meaningful way than a humanoid robot.

Is this a joke? An actually functional humanoid robot would be able to be dropped into basically any factory in the world and do a job that most humans agree sucks. And you think that is less impactful than building an expensive and very specific tool for a specific job that humans are already not doing very much?

> An actually functional humanoid robot would be able to be dropped into basically any factory in the world and do a job that most humans agree sucks

Oh really?

Can this Tesla bot pick up a strawberry without crushing it?

Can this Tesla bot identify a weed in a yard, and uproot the weed without harming the other plants?

------

Weeding and harvesting of fruits are just two, labor intensive human tasks today requiring the human touch.

There are also other human tasks: opening a door, walking up and down stairs, folding laundry. Etc. Etc. What can this TeslaBot do?

As I noted earlier: Boston Dynamic's 'Spot' is picking up landmines in Ukraine.

--------

But we know how agricultural robots look like.

https://youtu.be/8BAQCpUu3ww

Is humanoid even an advantage? No. It really isn't. Humanoid unnecessarily hinders your imagination to the possible efficiencies at play here.

Honestly, a weeding robot might just be a powerful laser + quadcopter drone. No reason for it to be human at all (bipedal, fingers, etc. Etc). Or maybe some chemical herbicide/poison applicator (spraying Hydrogen Peroxide or something like that?)

What the current TeslaBot can or cannot do is not the point I was making, I was merely responding to you assertion about the transformative potential of humanoid robots.

Humanoid is an advantage because all of our existing infrastructure is geared towards humans – hands with five fingers / two legs / etc. Again, that means a high-functioning humanoid robot could start doing shitty manual labor people don't want to do overnight.

This is obvious in a factory because you don't need to retrofit your entire factory, but even more obvious in a context like a home, where there are loads of small activities it would be nice to automate (gardening/washing/cleaning/etc) that would each require its own robot if you don't have a humanoid one, at which point... doesn't it make more sense to just have the one humanoid robot?

> What the current TeslaBot can or cannot do is not the point I was making

Yes it is.

> Humanoid is an advantage because all of our existing infrastructure is geared towards humans

Cool. So can Teslabot open a door? If not, when do you think Teslabot can open a door designed for humans to open?

> Again, that means a high-functioning humanoid robot could start doing shitty manual labor people don't want to do overnight.

Only after that humanoid robot can open the door and climb the stairs and reach the factory floor. Alternatively, you *design the factory around the robot* instead.

You know, so you can do things like...

https://www.youtube.com/watch?v=S8qkaTsr2_o

------------

Modern infrastructure, in the USA at least, is designed for machines, not humans. We have pallets to move objects around, forklifts to move those pallets, bays that directly lead onto trucks.

You don't design around humans. Humans can only lift maybe 50 Lbs of weight (25kg) safely/reliably without injuring themselves. A good pallet-jack can lift 5000lbs and be moved around the factory no problem.

Interfacing with the "machine" system of pallets and pallet jacks more quickly/efficiently is the meta. That's why an automatic palletizer / unpalletizer is a bigger deal than any humanoid robot Tesla can make.

https://www.youtube.com/watch?v=sXkEjbZtNic

The thing is of all the things Musk promised over the last decade literally nothing happened.

Remember the fully autonomous vehicle "by next year" in 2013 ?

The boring company solving your commute issues by 2020 ?

The 500 000 first fully autonomous tesla S of 2020 that were supposed to be sold to uber ?

The cyber truck which is "just around the corner" ?

This guy is selling dreams, I can guarantee you in 10 years we still won't have fully autonomous teslas, anything close to a proper robot or underground highways made by the boring company other than a few demos to milk the last investors who still believe in him

I agree that it’d be nice if he could deliver on more of his promises, but selling dreams is basically the job of upper management.

Also, most other companies promise really mundane and uninteresting things. Sure, they deliver, but it’s still not much fun.

> but selling dreams is basically the job of upper management.

And evaluating the dreams and funding them is the job of investors. Just following the hype betrays your own money, your own agency, and possibly makes the world a worse place. (The Hyperloop myth has only served to hamper high-speed rail adoption)

We all know this is a big advertisement for Tesla's stock, and maybe advertisements for workers to join Tesla. Now we can discuss if its worthwhile for shareholders and/or engineers to consider what was seen here.

> The Hyperloop myth has only served to hamper high-speed rail adoption

One guy wrote a badly sourced article claiming this and now its a proven fact.

Its total nonsense. Even if you believe that Hyperloop was some 4D chess against HDR (and it wasn't) then its still insane to suggest it has anything to do with the problems of HSR.

The HSR project started in the 90s and has been struggling the whole time. Blaming a few statements by Musk made in 2014 that 99% of the population no nothing about is delusional.

HSR funding wasn't cut because of Hyperloop. Literally no California government action happened because of Hyperloop.

I seriously think you are letting your dislike for Musk overwhelm basic logic.

Please, explain what Hyperloop has accomplished then. Its been 9 years since Musk proposed it in 2013.

You're latching onto a single phrase of my post and ignoring the problem here. Elon Musk proposes many things, some of which are harmful, in the hopes to gather investor money.

My point, is that investors (ie: shareholders) need to think about what to fund, or not to fund. Engineers who apply to his company, need to think about which teams to support and which teams to ignore.

I don't think that this Tesla Robot thing / Optimus has any hope of accomplishing anything useful. Disney / Boston Dynamics / Honda / etc. etc. have gotten far further with robot technology if anyone actually wants to join a competent team.

--------

If Hyperloop isn't weirdly bad enough for you, then Elon Musk's random obsession with Bitcoin and Dogecoin was also quite harmful.

> Please, explain what Hyperloop has accomplished then.

Nothing. I didn't make any argument that they hyperloop was good.

Not sure where I got the idea that I did.

> Elon Musk proposes many things, some of which are harmful, in the hopes to gather investor money.

He never gathered investors for the hyperloop. So not sure why you would bring it up as an example.

> I don't think that this Tesla Robot thing / Optimus has any hope of accomplishing anything useful.

Well, Tesla is a pretty big company, this project is a tiny part of their budget and is heavily based on projects they work on for other things as well.

And Tesla and Musk clearly disagree that this could be useful.

Again, please tell me what investor has Musk harmed.

I guess people might have bought some crypto currencies because musk likes them, but that their responsibility and those people are not investors. Musk never said those things will make money for them.

https://www.reuters.com/markets/us/elon-musk-258-billion-dog...

Don't just trust anything this guy says. Evaluate and think.

So by "investors" you meant "people who bought Dogecoin thinking it would be a good investment".

Personally that sounds more like Elon Musk has allegedly harmed gamblers.

By "investors", I mean Elon Musk fanbois who got burned listening to his tweets on Dogecoin.

The same fanbois who are about to get burned on this Optimus robot and Tesla stock, IMO anyway. This is all a huge waste of money for the car company.

The same fanbois / Tesla investors who got $Billion of dollars of Tesla's investment money trapped in BTC at the height of the Bitcoin bubble.

https://www.cnbc.com/2022/07/20/tesla-converted-75percent-of...

SPO -> $Billion gathered from Tesla shareholders / investors, that money went into BTC and WAS GAMBLED by the technoking. But no biggie, they'll just SPO again and get more money from the next set of share buyers.

Were you also saying Tesla fan were stupid when Tesla invested in its own manufacturing when everybody said 'Just go to Magma'. And then again when they made their own batteries?

Overall Tesla strategy has been very good and it has made most people money.

Its just your opinion that this has now changed.

Yes, autopilot/FSD isn't there yet. Otherwise, Tesla is on a great track and growing like mad. How is he over promising with Tesla? Yes, Semi/Cybertruck are late, but it seems they are about to come to market. Meanwhile the Model Y is growing like mad, might have delayed the new models.
Yep. Seems like all auto manufacturers are struggling with supply chain management, causing delays. Most manufacturers don't talk about future products at all.
They often do, and the wait might be longer than with Tesla. Just look how long the ID.Buzz has been teased until it came onto the road. They also constantly show prototypes at trade fairs which never make it into production at all or only significantly changed (usually much more boring).
You have a loose definition of "all" and "literally nothing."

I don't expect entrepreneurs to be 100% successful in every venture and the things he has delivered on are pretty substantial.

> The boring company solving your commute issues by 2020 ?

Where when? Please show a source. I have followed Boring pretty closely and that was never claim.

> The cyber truck which is "just around the corner" ?

The Cyber truck is certainty delayed, no question, but that's not that crazy of a thing.

The have invested billions in the production facilitates for the product and there is every indication that this is product will be released.

If Tesla was not making progress at all this would be a major issue, but they are growing fast still. So they managed to grow the company even without the Cybertruck. That seems to be a good thing given the potential the company still has with that truck.

> /nderground highways made by the boring company other than a few demos to milk the last investors

Boring company is still in operation and has number of projects. Has any investor lost money? And Musk himself is the largest investor.

What investors have lost money in Musk companies? What investors has he dis-pointed. You seem to make an impassioned argument that investors are being milked dry by Musk. But who? Tesla and SpaceX certainty have made investors happy. Boring hasn't lost money from anything I have heard.

What a stupid article. To be honest I was impressed considering this is less than 1 year of work.
It was experimental unveiling for attracting talent. The article is more "cringey".
People are missing the point. The robot is a plataform to evaluate the generalization of their new ML architecture. The benchmark shouldn't be BD or Honda, but Google and Disney who have shown (non humanoid) robots doing similar things.

Of course, it is ok to discuss the mechatronics, but as many pointed out, making them is a solved problem.

> “None of this is cutting edge,” tweeted robotics expert Cynthia Yeung. “Hire some PhDs and go to some robotics conferences Tesla.”

And this is exactly the problem created by social media.

I don’t care for Musk or any other tech celebrity, visionary, billionaire, or however you want to describe them.

But why does some demo have to be so extremely impressive? How is this newsworthy?

The real problem here is social media. Every toxic, awful loon is out getting their negativity and grievances amplified 1000x.

The biggest element I think people are underestimating is Tesla's level of vertical integration in 2022. Boston Dynamics has no large scale production facilities to speak of. They have not invested millions in ML hardware/software/training architecture. They have not developed their own electric actuators designed to be produced cheaply and at scale (in factories they already own). They are not sourcing their own lithium for batteries. etc. etc.

A big reason Tesla is entering this space is because a huge amount of the things needed to make a functional android that can work in a factory they are/were already doing. There are so many synergies it would be stupid for them not to try something like this. And this is a decent amount of progress for less than a year, to the point where I'd be surprised if this didn't cause some BD people to swap teams.

And BD being "ahead" isn't the dig people seem to think it is. BD being ahead just means there is a benchmark Tesla knows they can get to with current technology and (probably) surpass. Saying BD is ahead is like saying Lockheed Martin/Boeing were ahead of SpaceX in 2009 – trivially true but woefully shortsighted.

Actual robotics experts seem to think that the design, particularly the actuators, is fundamentally flawed and cut some corners that won't be particularly easy to un-cut on the way to an actual product: https://twitter.com/_brohrer_/status/1576368293719736320
I dunno, SpaceX's Starship was gonna be composite/carbon fiber and then they decided "actually maybe stainless steel is better" and went with that. Which in my humble opinion dwarf's "our actuators need to be better/different" as an engineering decision.

If there is one thing Musk's companies seem to have it is a combination of both velocity but also a willingness to go against the resulting momentum if it seems like that velocity is in the wrong direction.