I find this to be surprisingly significant news for Tesla. The thesis until now has been that Waymo is expensive and Tesla is cheap. Now Waymo is starting to close that cost gap. The narrative has been that Waymo is just a fly that Tesla can squash whenever it wants, but there is very little evidence to support that.
In the long run, Tesla also faces the challenge that the pace of vehicle development has slowed significantly; competitors are evolving faster, and Waymo can pick the winning horse and switch if necessary.
Muskâs interest seems to be shifting elsewhere, and it feels like âcars are boringâ to him. I wonder what is propping up the share price.
Iâve been driving a Model Y for about three years now and consider it a very good car, but development has slowed down tremendously while competitors are accelerating.
When it comes to autonomous cars, the question has always been which will happen faster: Waymo/Nvidia etc. scaling the price point of LIDAR low enough, or Tesla solving vision-only autonomy.
The first problem is purely techno-economic, where itâs just a matter of how many quarters/years it will take; we have already seen how LIDAR prices have practically collapsed in just a few years.
The latter is a binary either/or problem, meaning we cannot know if Tesla or anyone else will ever solve it (i.e., whether Level 5 autonomy will be achieved).
I think itâs a poor decision for Musk to stubbornly stick to a camera-based system. As mentioned, the price difference of sensors will eventually become insignificant. Furthermore, even if the camera system is made to work better than the acceptable minimum, it will always be lagging behind solutions that have multiple âsensesâ available. Itâs like trying to program a human brain where one is missing the sense of hearing. No matter how good the software is made without hearing, over time, the one programming with the sense of hearing will always have the advantage.
âAfter a thorough review and assessment of the technical documentation, the Danish Road Traffic Authority agrees with the RDWâs assessment that the system will contribute positively to road safety by assisting the driver while driving. The Danish Road Traffic Authority therefore accepts the provisional type approval of FSD Supervised.â
First data from the Netherlands has been published:
Has anyone looked into this: a large portion of accidents happen to the elderly, new drivers, and people driving under the influence. Few of these people drive a Tesla equipped with FSD. So, in these â3.5x saferâ comparisons, are we comparing apples to oranges or apples to apples? Additionally, are personal injuries being compared to cars of the same age, or are old, unsafe cars included as well? Is there data showing how many crashes would have occurred if the driver hadnât intervenedâmeaning, how often does FSD make a wrong move?
Edit: Has FSD managed to eliminate phantom braking that endangers road safety?
This was just discussed in the thread a moment ago. Teslaâs figures are not comparable with any other data, and these â3.5x saferâ claims of theirs are absolute horse-shit.
The image had an explanation for thisâmeaning Teslaâs FSD driving was compared to Teslaâs normal driving. Here it is enlarged.
Personally, I believe that middle-aged people suffering from sleep apnea, the elderly, enthusiastic new drivers, and drivers tired for various reasons are the first ones who want to use FSD. Therefore, FSD improves safety, as all above-average drivers can handle the wheel safely in Toyotas and VWs and donât need to worry about FSD Teslas. This is what the RDW, Danish authorities, and others have stated in their reports.
Phantom brakings are usually not emergency brakings but unnecessary normal decelerations where Tesla tries to prepare for a potential problem situation. Not all of these annoyances have been eliminated from FSD, but I havenât seen reports of crashes caused by this. The emergency braking statistics are also in the image, and there are far fewer of these with FSD than normal.
I strongly disagree. New drivers cannot afford a Tesla with FSD. Seniors donât choose Tesla because itâs too much of a âcomputer on wheelsââtoo high-tech and lacks physical buttons. Even among substance users, the vast majority do not drive a Tesla equipped with FSD.
Phantom braking is not the same as emergency braking, but it does compromise road safety. It has happened many times while riding in a Tesla where it slows down quite abruptly from 120 km/h to 80 km/h on the highway.
