Quote:
Originally Posted by Brupal
Interesting that you mentioned costs, Bizaro86. Yes, Waymo currently offers driverless ride sharing, I’m actually envious that you have had the chance to experience it, sadly I have not. We can debate whether using cameras and neural net software is better than cameras and lidar and radar in a different post but when it comes to cost Waymo is at a disadvantage. While it’s true lidar sensors are not overly expensive what happens when you have a conflict between what your lidar sensor says and your camera says? Quick, lives are on the line. You can’t wait, you have to go with one or the other. Which is it going to be? Click - you go with the camera because it sees what is in front of you. That is why Tesla chose to go with camera only driving, because too many inputs can lead to distortion and you have to go with one. But setting the technology aside, currently it costs Tesla approximately $30,000 to make a Model Y which comes equipped with FSD software which Tesla can turn on for free and put that vehicle out into the fleet and start earning robotaxi fares. Whereas Waymo, who doesn’t manufacture its own vehicles, has to buy a Jaguar iPace for about $90,000 (and sadly for Waymo Jaguar no longer makes the iPace), and then Waymo has to buy 13 cameras, 4 lidar sensors and 6 radar sensors, and then it has to retrofit them to the Jaguars (that it didn’t build) before it can add that vehicle to the fleet. Such a platform is not easily scalable. Waymo has approximately 2000 vehicles currently capable of operating as robotaxis. Every model Y that rolls off the assembly line (about one million per year) is ready to go as a robotaxi. Tesla will be able to roll out robotaxis into many cities quickly - much faster than Waymo, and capture market share before the competition can. And given the significant cost advantage, Tesla could charge far less than whatever Waymo chooses to charge, and still do so profitably. I hope Waymo and other companies are able to provide autonomous ride sharing and self driving vehicles, but currently likely only Tesla has the ability to do so profitably and that means they will likely succeed while others may not. We shall see. But along the way our roads will be safer for all of us as fewer impaired and just generally bad drivers are out there.
|
I mean, Waymo has switched to Hyundai Kona EVs going forward (no prize for guessing if that's cheaper than the Jaguar). Building cars is absolutely a commodity business, and having the ability to do so isn't a competitive advantage. Byd is eating Tesla's lunch in China and has started taking over the Indian market as well on the physical car side.
In terms of sensors, the reason to have more than 1 type is for when one is wrong. Then you take the more accurate option (which is Lidar's 3d picture not a camera's 2d picture) or make the more conservative choice. Waymo obviously has this figured out - their cars are driving around San Francisco (which is a pretty difficult environment imo, I wouldn't really want to drive there) more successfully than humans.
I'm not saying Tesla's system is bad, but the problem is it has to be perfect. Last time I was in SF a friend drove me from my hotel near Union Square to the airport. He has a model Y with self driving, and he put the destination in and it drove the whole way. Except for once where it froze and needed driver input. True self driving can't be 99% good, it has to be 99.99999% good - and with camera's only I think the amount of human intervention required will always be too high