Is Tesla’s Full Self Driving dangerous?
We have seen on the news that some people think Tesla is using public roads and not professional drivers to test their system putting everybody else at risk.
Life is dangerous. We get out of our houses every morning thinking that nothing is going to happen to us. But getting out to the streets always has risks. Driving increases that risk in a significant manner as it is one of the statistically most dangerous activities we do every day.
When we take our car, we think we are good drivers and nothing is going to happen to us, but most drivers have already been involved in some kind of car accident and many more will in the future if nothing changes. Cars are safer now as you can tell from casualties data but the figure is still too high for our society to accept it.
I think self-driving technology is here to save lives. Probably, in the meantime, they will make our lives easier and car manufacturers will profit from it. But, in my opinion, the main reason they are here is to lower the casualties figure.
FUD on Tesla
After watching the CNBC video where so many failures were exposed about Full Self Driving beta, I want to overview what self-driving will depart us in the near future, and what can we expect today.
Although legally FSD beta is just a level 2 autonomy feature, people who have been watching YouTube’s videos from numerous fans going from one place to another know that it is more.
Of course, there are still many errors and you cannot rely on the system fully. But if you’ve been following the course of the incremental updates Tesla has been releasing, the system has got better and better.
I agree that you cannot still relax back on your seat because disengagements are sudden and you have to take control immediately. Also, sometimes, the car is not making good decisions and if you are not fully alert, it could cause an accident.
FSD beta today
Right now, FSD is not for everybody. You can mildly relax while driving on the highway as the system will maintain speed, get curves comfortably and pass other cars while smoothly changing lanes. This part is 99% ready and you might have a warning or a disengagement once in a while.
But city streets are a completely different story. Forget relaxing here. It is amazing how the system works, the decisions it has to make, and the smoothness of most maneuvers, but you could drive yourself in a more relaxed manner than FSD does.
The software is still in beta and, by accepting it, you become a test subject for Tesla to improve and you are exposed to some risks. You know that right? Then, a video showing how the system fails is like a new Windows version beta tester showing how his system crashes. Sure Windows won’t hurt you and your Tesla could, but no serious injuries have been reported up to now on the FSD beta program that, at the end of January 2021, had more than 60,000 beta testers.
The idea is that Tesla will use this data to improve the system and that these improvements will be a lot more significant and faster thanks to a large number of testers. The normal approach based on hiring a couple of hundred professional drivers to test the system is not fast enough and doesn’t gather enough data to match the Tesla way of doing things.
Their methodology works in a similar way as the pharmaceutical industry. They start trials with a small set of subjects and, later on, they do it with a larger number of volunteers. The main difference is that Tesla is not being supervised by any governmental agency before starting their wide testing phase.
FSD in a few years
FSD is updating once or twice a month improving the system incrementally. Beta testers agree that most of these updates improve significantly the experience and smoothness of the trip and that, in just a few months, the software got so much better.
Although Elon Musk told us in the January investor’s call that he would be very surprised if autonomy is not solved by the end of this year (2022), we all know how Tesla’s CEO measures time frames. He has been promising autonomy since 2016 and we are not there yet.
I think we should not be very ambitious with this technology in a short time.
The system will continue improving and I would expect these phases:
- By the end of this year (2022) we could have cars that are able to drive themselves on the highway with 99,99 % confidence, but drivers should still be alert, facing the road, and ready to take over at any time. City streets driving will still be in beta for the full year 2022.
- In a couple of years, 2024–2025, we might have cars that will be completely autonomous on the highways and, if regulators allow it, we could start doing other things in the driver’s seat as reading once we can take the sight out of the road. The car should be able to alert us a few seconds in advance or stop the car in a safe area if there is a situation it cannot handle. City street driving should improve but still would require full driver‘s attention.
- By the end of the decade, I expect the problem to be solved but we will still need a human driver at all times in the car for consumer vehicles. Full autonomy without a driver might only be implemented in commercial vehicles such as ride-hailing or freight transport.
By the end of the decade, I expect autonomy to be good enough to reduce the risk of injury while driving by at least 10 times, as human errors, which are the most common cause of accidents will be almost completely erased.
This will only improve as more old vehicles are replaced with new more capable ones. This process will take at least 10 to 15 years more.
Most drivers will be happy enough if they can use their commute or travel time to chat with their family, read a book or browse the web instead of staring at the road. If this also brings improved safety, the future will be for sure better.
If you liked this article, you can read more on self-driving technology here: