Hacker News Reader

Comment by ddxv | Hacker News Reader

I love that this is landing in Austin before the cybercab. I think it will really drive the competition for Tesla to put out an (actual) self driving car. Or maybe they pivot to the next hype train.

View on HN

The problem is neither Waymo or Uber make their own cars, and the cars they use weren't purpose built for self driving. Both of those points give Tesla a huge economic and competitive advantage in the long run. Uber/Waymo are the middlemen, once OEMs start building their own purpose built self driving vehicles there's a good chance Waymo/Uber get factored out of the equation.

Not if OEMs don't have the technology. Making cars is the easiest part of self driving. Once you have the tech, you can farm the manufacturing out to many OEMs (like Waymo is doing with Hyundai and Geely) and then OEMs become the middlemen.

If Waymo continues to grow their OEM partner pool, then presumably they'd have a better ability to put cars on the road than Tesla. I'd say right now it's "advantage Tesla" but could easily see this flipped on its head if Waymo announces more partnerships soon. Could very easily see them partnering with major US automakers like Ford or Stellantis (lol maybe even GM, but that seems extreme)

> I'd say right now it's "advantage Tesla"

Tesla self-driving doesn't exist and will only work on Tesla vehicles if they ever develop the capability.

Waymo self-driving works now and has been demonstrated on multiple vehicles. Waymo could utilize the manufacturing capacity of the entire auto-industry.

Oh yeah, I totally agree. Should have specified. That advantage only applies to making vehicles. In terms of driving, operation, and pretty much everything else, clearly 'advantage Waymo'

What’s the advantage of owning the full stack? Some margin on the base machine? Is that that huge a factor?

Ask Apple?

But the point there is that by controlling the stack Apple can improve the overall UX, from encryption to battery life or what have you. What’s the equivalent in self-driving cars? Better handling?

Perhaps. More broadly, reduced constraints. Written differently, being able to choose the layer at which you solve the problem.

An example of this is that in robotics, the ability to properly insert a piston into an engine block without damage to the piston or block was an unsolved problem for some time in robotics. Doing so using machine vision and robotics control theory required engineering constraints that resulted in a too slow and expensive a mechanism for mechanized assembly lines.

Because of the total control of the assembly line, this problem was solved in the physical space through a reactive physical assembly using two connected planes (I'm sorry but I'm having a hell of a time tracking down a reference) rather than in software.

Apple doesn’t own any factories to make their hardware. Their vertical integration goes as far as hardware design and software.

Pretty much exactly what companies like Waymo are trying to do.

It's fair to say that Tesla's vertical integration is even deeper. Might Microsoft Windows or Google Android be a better example than Apple? Does Waymo actually specify the entire vehicle (i.e. have it's own engine design) or just the systems added to the vehicle?

Indeed, Windows or Android are a better example than Apple. Waymo only designs the systems added to the vehicle, but they do have input over some parts of the vehicle design. Geely (Zeekr) is building a purpose-built robotaxi for them that's based on their SEA-M platform, but with Waymo customizations.

I'm not sure Tesla level vertical integration is really necessary.

Cars are commodity. The brains/driver is the value.

Well seeing that Tesla self driving cars are all smoke and mirrors and has been for years and Waymo os operating now…

Well at the very least it will be interesting to see the fallout when the lidar-less Tesla robotaxi flattens its first pedestrian.

Waymo published details of a new model recently that only uses cameras.

https://waymo.com/research/emma

Or an earlier one. Been a while since we heard any progress on how Hyperloop will own the future.

I don't want to be the one sitting in a Waymo when it decides to clusterfuck up traffic.

In SF, when Cruise & Waymo were both operating at the same time, there was a lot of animosity towards autonomous vehicle driving behavior (especially emergency vehicles). Both companies got lumped in together, but this seemed to stop as soon as Cruise paused operations.

It's nice that Austin will get a chance to see Waymo first, and have the ability to differentiate between the performance whenever Tesla launches. It'd be great if both performed flawlessly, but I have my doubts about one of these companies ability to act responsibly...

Austin already had Cruise until Cruise operations shut down in late 2023.

I tried Cruise in Austin, and it was nothing more than a curiosity. What should have been a ten minute ride took almost forty minutes because the car took a circuitous route with lots of backtracking, presumably because the cars were programmed to stay away from certain intersections and avoid left turns. It was my first time in a self-driving car, so the cool factor was definitely there, but it was far from being a real product.

Not in any meaningful way though. It was limited to only night time and select riders for most people to even notice.

Both companies got lumped together, but Waymo had and continued to have issues after Cruise stopped operations. Some of those continued making headlines, like the 3am honking.

During the time they were both operating public service in SF, Waymo had slightly fewer incidents proportional to their fleet size, but almost all of them were extremely minor. Cruise had mostly minor incidents with a few catastrophic failures.

Elon has already pivoted to rockets, space internet, social media and dismantling governments, if you are still holding Tesla in my humble opinion you are already holding the bag.

Reply on news.ycombinator.com