Hacker News new | ask | show | jobs
by Mobius01 1356 days ago
The Tesla Robot was underwhelming. I appreciate the engineers on stage highlighting their work, but it all felt a generation behind Boston Dynamic’s efforts.
8 comments

In a way. The motion was not spectacular like Boston Dynamics.

But any walking robot especially at that height and size is a significant engineering achievement.

And they have been working on it for months rather than years. And the system is entirely electric unlike Boston Dynamics.

The last demo was a person in a suit.

Your expectations were unrealistic if you are not impressed.

Similar robots are built every year in colleges around the globe by jr robotics teams. Walking robots have been around for 25 plus years. There was zero to be impressed about. Walking robots are to robotics as hello world is to programming.
It was not simply programmed to walk, it was also using computer vision to decide where to walk too
It looked like a very crude depth occupancy model.
>very crude depth occupancy model

Same as Tesla's FSD and 100% vision based. Is it crude too? I thought it was state of the art…

"Walking robots are to robotics as hello world is to programming."

You say this confidently, but it is nowhere close to true.

Can you show us some examples?
Atlas is also electric. An electric motor tied to a gear pump presumably.
I mean it was an embarrassingly janky piece of shit that looked like a slight breeze might send it to the ground.

Nobody would first look at that and then want to get into a self driving car built by the same company.

I mean.. checking out twitter, there are just hundreds of people, at least, who were just blown away by this janky nonsense and the "fact" that tesla will "ship millions" of these this year "for $20k."

There's some kind of mass delusion going on.

In reality its going to be 10+ years and it'll be more than 20k. Thats assuming they actually release it, they seem to be collecting a pretty large list of vaporware right now. Cybertruck and Semi for example which seem to be stuck in development hell.

They seem to have a real issue actually shipping. Partly because Elon is incapable of understanding realistic timeframes.

Unrealistic timeframes also has the benefit of getting the most out of people. But Elon is getting quite a bit of flak for the expectations it sets.

It sure will be interesting to see where FSD & Teslabot is in a couple of years. The pace of innovation is extremely high at Tesla and should make you wonder what Boston Dynamics, Waymo and others have been up to for all these years.

Making things that actually work outside of imaginationland?

How can you look at what they showed yesterday after the rest of the world has made things 109x more capable and conclude that they are innovators in the space?

Even the rando Chinese companies have done a better job cloning spot mini than this.

More things working outside of imaginationland than most produce in several lifetimes.
> tesla will "ship millions" of these this year

Just like million robo taxis in 2020 :-)

I think he said like 3-5 years.
I look at that and I want a self driving car by Tesla. Did you see the rest of the presentation about their self-driving tech?

This was amazing by this with the short time it's been in development, and the vision for this project is beyond anything that has been done with humanoid robots. Sure they might fail, but it's an exciting venture to follow.

Have they driven from LA to NYC as Elon promises they would by 2019?

Are Model S owners making more money than they are paying in leases thanks to lending their cars as auto cabs like Elon promised they would by 2020?

Is hyper loop even a thing or just something Elon admitted was something he came up with to try and kill public transport alternatives?

The Boston Dynamics Robot is far more of a commercial reality than the CyberTruck if you really want to put things in context.

> Have they driven from LA to NYC as Elon promises they would by 2019?

Why do people only ever compare Tesla efforts to Elon expecations.

How many other companies have cars driving around in pretty much every environment on north America? Their competitors have a totally different approach and limit themselves to a few fixed locations that they map out in extreme detail first.

If you take all self driving cars in the world and put them at random locations in the US and tell them to drive somewhere I don't think any company will do better much better then Tesla.

I'm not really a fan of self driving but just clowning Tesla for not reaching the goals set by Elon isn't really interesting or insightful.

> Is hyper loop even a thing or just something Elon admitted was something he came up with to try and kill public transport alternatives?

Hyper loop was never a product. Musk even explicitly stated that it wouldn't be a product anytime soon. He explicitly state they did some evaluation on it and would release an engineering blue paper. This was literally stated outright at the time.

I really don't get the logic 'I wont do X but I find X interesting' and the response a few years later is 'Fuck you, you promised to do X and it didn't happen'. Like that makes people sound insane.

And he never said it was to kill public transport. That was a single news article where a journalist took Musk words out of context to spin his own story.

And people who think California High Speed Rail is failing because of Musk is fucking delusional anyway.

> The Boston Dynamics Robot is far more of a commercial reality than the CyberTruck if you really want to put things in context.

That is totally absurd comparison that makes no sense and illustrates nothing.

Yeah for sure, the autopilot claims and did not come to fruition, for sure, lots of misses when Elon thinks something happens and when it actually will. I don't think personally he's intentionally deceiving the masses.

Not really sure about the hyperloop / train thingy, need to read up more on it.

And BD's robot is not a commercial reality, how many have sold (if we're talking about Atlas and not the dog)? The Cybertruck will start production next year, sure it's behind initial time-schedule but it will be produced and sold in high numbers.

It's pretty hard to get excited for self driving from Tesla. So many of Musk's claims (i.e almost all of them) have turned out to be complete bullshit. They're seemingly still years away from full autonomy, despite him constantly trying to make people think its coming any day now.
BD took ages to be were they are right now.

I actually think Tesla showed more than I expected.

At least we now start the rat race for robots.

That rat race started like 30 (40?) years ago in Japan.
Probably true but I do mean the real progress part.

Up and running android, big money, fast development.

It was underwhelming, but it's to be expected for 1 year of work.

This is also an electric motor robot vs pneumatic actuators from BD.

Choosing the wrong technology and failing to make it work is not an accomplishment.
Are you an expert in robotics? Why is electronic actuators worse then hydrolics?

In most other places we are switching away from hydroponics because electronics are getting so advanced.

Electronic actuators is defiantly cheaper and simpler so for a consumer product its seems to be the right choice.

hold my gigapress
Why are electric motors better?

Excavators etc use hydrolics, presumably for a reason.

I suppose if you want dexterity? But then you'd lose load carrying?

Did you see the part where the single Tesla actuator was lifting the piano up and down?
Yes but if you're statically holding it that's going to burn out motors etc? And just because it can, doesn't mean it's better.

But anyway it was just a general question. It doesn't seem self evident to me that electric motors are 'better' so I was just wanting someone to chime in and explain.

Yes, a bit underwhelming, but it's a recruitment event and you've got to start somewhere. Question is how fast and if they can catch up with BD in terms of control.
I assume they’ll catchup as fast as they catch up with their FSD to Waymo. Which is driving customers for years, while Tesla’s FSD cannot go without intervention for more than few miles, at best, in any more complex environment. And yes, I know Waymo uses HD maps and limited geo area. So does Tesla, and they’re explicitly mapping and testing areas driven by popular influencers.

And it’s assuming that there’s actually a market for humanoid robots. Other than futurology folks getting excited about that form factor, there’s no real market for it, and research shows that humanoids robots are almost never a most optimal answer.

> So does Tesla, and they’re explicitly mapping and testing areas driven by popular influencers.

This seems like a nonsensical swing at Tesla. They have literally 100000s of people on the road with FSD in all of the US.

Claiming that its a few influences is just outright lying.

You are also totally wrong in your understanding of how Tesla FSD works. They are not mapping those areas, save those maps and make them available to other cars. That is explicitly not what they are doing.

More accurate to say that they are building a training set with the input from 100000s of cars driving around literally all over the US.

If you mean catch up as in FSD is making billions while Waymo is losing billions, yes the comparison is apt. Boston Dynamics tech is very impressive but they don't seem very good at commercializing it. Tesla bot may be janky now, but I could imagine it could do some useful tasks in their factories in 5-8 years or so. We'll see, I think the development will take even longer than self-driving, but at least it will be fun to follow the progress if they're as open as with FSD
Tesla FSD isn't making billions. What are you on?
It is. Let's check by two measures:

1. 160k people have bought and recieved FSD Beta at somewhere between $5 and $15k (today's price). Let's say an average of $8k, that means $1.28 billion. More have bought it but are waiting to get access.

2. FSD take rate is estimated as of now to be around 7% of vehicles sold world-wide. At the current price of $15k and the estimated 1.4 million vehicles sold this year, that amounts to $1.47 billion.

Numbers are rough estimates, 50% of revenue is held in reserve until FSD is out of beta, and many haven't received the beta yet, still by any measure they are making and will make billions. What you personally think of the product, or if you think Elon is a "snake oil salesman" is not really relevant.

To strong arm any counter argument, it's possible that all these buyers will realize it actually sucks/doesn't work, sue the company and the whole thing falls apart, but I don't see any indication of it. There are plenty of unhappy and happy customers with FSD, but they've clearly found something that people are willing to pay a lot for and many are happy with it. Clearly it needs to improve much further in order to sell this to the wider public beyond people who think it's fun to babysit an AI.

The product Tesla has sold for money is incomplete with no credible projected date of completion as the general problem of L4/L5 driving has never been solved before. Any profit from pre-selling a solution to a never-before-solved problem which requires an unknown amount of future work to make work as advertised absolutely does not count as “making billions”.
technically that's still not billions.

It's still crazy to me that people are paying all this money now for a product that still isn't delivering the promises of 5 years ago though.

Are you including the amount they're losing by having to recall those vehicles with FSD? Or the fact that there are lawsuits over how FSD doesn't work? Or any number of accidents caused by FSD being clearly buggy?

Like ignoring how poor your breakdown is over it making 'billions' they've been hit by wave after wave of bad PR. Like the various videos and lawsuits over how poor their quality control is and how many lies they've peddled over FSD.

It seems strange that this comment was so aggressively down voted. Anything related to Elon gets the same treatment by HN that I feel that Elon invites a visceral rather than an objective reaction.
Because Elon is snakes oil salesman, who thinks he’s smartest and most important human in the history. He also happens to have some very successful products, that his fan base uses to silence any criticism.
If snake-oil salesmanship is all it takes to land rockets vertically and single-handily kickstart the transition to EVs I'm all for it. We need more snake-oil salesmen leading corporate America!
There can't be any objective discussions about Elon any more =) John Carmack finds Elon to be smart and really knowledgeable (listen to the latest podcast from Lex Fridman). And you would think HN trusts the words from Carmack? But most think Carmack, who knows Elon personally is wrong, and their opinion based on some random internet articles, is right.
Elon is super smart. He’s also a snake oil salesman.

These are not contradictory statements.

Unfortunately he chose his intelligence to enrich himself by fooling others. He could have chosen other options. It looked like that was the direction he was going in. By trying to help people. But it’s clear at some point he realized he can make far more money claiming to do something than by actually doing something and decided to focus on that.

Exactly. He lands rockets so he cannot lie? Is that a new law of universe?
So he is at the same time a snake oil salesmen that also not a snake oil salesmen because he revolutionized 2 major capital intensive industries.

You realize that makes no sense?

People just can't give someone any credit if they don't like the person or they have slightly different political views, I guess it's just the times we live in, zero tolerance.
Anyone who uses the word “snake oil salesman” to describe Elon Musk is delusional and has gone way too deep down the anti-Elon rabbit hole.
Do you at least get paid for this?
Kind of actually. 10x’ed my net worth since 2019 with TSLA stock. It pays when you can see through the BS fud like your statements.
He's generally a force of good for the world but can come over as a bit of a dick sometimes, there are much worse people out there if you need to hate on someone.
I don’t know. Richest person in the world, whose ego is only thing bigger than his wealth, is pretty good person to direct at least some hate at.

And while some of his lies are just to kill competition/extract money from the rich (like FSD), but some are just disgusting.

With neuralink he was openly telling people for few years that his implants will solve their health issues. Implants that killed most of the monkeys in the labs.

It’s disgusting to lie to people with really serious health conditions, give them false hope and likely have them reject other treatments, because Elon said he’ll fix them. You don’t think it’s a good reason to pour some hate at him?

> With neuralink he was openly telling people for few years that his implants will solve their health issues.

He has invested mostly his own money in that.

Are you claiming there is no medical use what so ever for the technology they are developing?

Based on what knowlage do you make that claim?

> Implants that killed most of the monkeys in the labs.

What data is that based on?

> It’s disgusting to lie to people with really serious health conditions, give them false hope and likely have them reject other treatments, because Elon said he’ll fix them.

What? I have heard most of Neurolinks presentation and not a single time did Elon ever suggest that people just delay other treatments. Neurolink has clearly mostly done animal trials and product demonstrations.

I at least have never heard Musk claim that Neurolink would have a product on the market within a short time period.

Please provide sources for these claims?

I used to think like this, then I looked at myself.
I need someone better to dickride. Got any options?
Here's a good thread about Optimus vs. Atlas: https://twitter.com/EzerRatchaga/status/1576140275847610369
How many Boston Dynamics robots have been made and what's the price? And how long ago did they start?

Tesla is going for volume and just started.

Before we even mention Boston Dynamics, can we talk about how it compares to the Honda Asimo?

Because it doesn’t.