My reaction to the Tesla robot after working on robots for a decade | Hacker News Reader

My reaction to the Tesla robot after working on robots for a decade

youtube.com

16 pointsrexstjohn17 comments

Human walking is just us falling forward and then subconsciously correcting it - research found that our footsteps are never the same and can vary by up to 80%.

It seems like a hard problem this human locomotion - almost like self driving cars.

Worth watching on 1.5x speed. Good thoughts and info on directions for future robotics.

Not very surprising that a company with a lot of experience in building electric motors goes for servos.

Doesn't necessarily make it a right choice, though.

However, Tesla (usually) does the right thing after exhaustively trying all the wrong ones first. :-)

“servos are the local maxima for robotics“

“The future is tendon driven and evolutionary algorithms “

“passive dynamics”

You realize both rocket and car manufacturing experts said the same type of things when Elon entered their fields. He can’t do this, can’t do that, should do this, should do that… etc

Let’s see how things progress 5 years from now

This seems to closer to hyperloop than to cars (and even there I wouldn't be that surprised if Cybertruck is not shipping in 5 years)

Yes. I remember how they said "he cant do self driving, people will die", and then he did self driving and people lost their heads. Plus the Lidar thing and plowing into huge obvious stationary objects.

Tesla robot moves like Asimo, very static "I have poop in my pants" walk. Little inverse kinematics, mostly looks like preprogrammed/pre learned moves that work only on known flat surface. Boston Dynamics Atlas it is not.

We also remember the claims teslas were going to pay themselves back as robotaxis, gigapress wasn't going to be producing 60+% scrap, tesla motors and controllers were special sauce that no other company could come near (lucid says hi), not to mention cybertrucks, telsa semi and tesla roadsters being vapourware...

You are reading way too much TSLAQ bullshit.

how is it bullshit?

I'm impartial - I don't have any tesla stocks or own one (have only taken one for a drive).

They are OK cars, but musk is a bullshitter and overhypes a lot, and the cracks in the facade are starting to show as other manufactures release EVs.

The 60% scrap on gigapress FUD only comes from TSLAQ. This type of BS has been going on for years against Tesla. They take a new innovation that is being tested, which of course is not going to be optimal at first, then blow it up and make up shit (the scrap story came about from TSLAQ looking at some photos outside of the plant, then literally making up huge scrap percentages from pseudoscientific deduction. Similar stories go back for nearly a decade, same BS all the time.

I mean its reported tesla are looking around for alternate press suppliers and people in the industry have also commented that trying to make a casting that big is pretty ambitious and a 60% scrap rate is not unexpected.

People have also calculated deliveries of gigapress cars to confirm the scrap rate so its not BS.

Then there is the issue of the casting cracking in an accident which can potentially write off a car that would have been repairable with traditional spot welded designs, this will increase insurance rates for Teslas long term.

So it remains to be see if its a genius move from Musk, or if traditional spot welded designs are in fact the better approach.

So it remains to be see if its a genius move from Musk, or if traditional spot welded designs are in fact the better approach.

Fair enough. There is a chance it may not work as they planned, but the fact that they are buying more gigapresses tells me they are confident they will get it to work.

Munro and associates did a couple of videos on gigcasting. They did talk about what happens in crashes and did not see it as a major concern. Munro is universally recognized as the leading experts in automotive manufacturing.

link to video where they talk about cracking the gigacast and why it doesn't matter:

https://youtu.be/WNWYk4DdT_E?t=335

Watch the full episodes here:

https://www.youtube.com/watch?v=WNWYk4DdT_E

https://www.youtube.com/watch?v=dyde8G7mp-4

Sandy has his own agendas - since he started to buy tesla stock I take his commentary with a grain of salt.

You notice when he talked about a failed casting he didn't mention anything around the fact that castings are brittle relative to sheet metal and bits can snap off vs flex.

So something like a suspension mounting point can break off the casting and that's a write off as its very hard if not impossible to weld back.

Actually Munro has sold all his TSLA stock, which wasn’t much to begin with, precisely to address any criticism about being biased.

Munro actually started off being severely critical about Tesla’s build quality, but once they started tearing down the vehicle, started realizing the superior manufacturing prowess of Tesla, they switched focus and bet their company on electric vehicles being the future.

I’ve been following Munro for over a decade, I’m involved in the automotive sector, nothing to do with Tesla, I’ve seen Munro’s full evolution and 180 turnaround on Tesla.

Musk's advertising strategy with Tesla has been to keep telling a bigger lie. Telsa's full self-driving is neither full nor self. FSD doesn't work right. But that doesn't matter because the lie sure works. People keep buying in even after 9 years of false promises:

https://jalopnik.com/elon-musk-promises-full-self-driving-ne...

The Tesla bot is just the next lie. It doesn't matter if it never works right. It doesn't matter if it is perpetually "one year away". People will buy in, just like they did with FSD.

make passive dynamics great again!!!

Reply on news.ycombinator.com