I knew it would have fun and useful assist features, but I didnt ever think it would get to level5 with its current sensors.
never bothered me. like I said, its a very fast car, it handles well, its style is acceptable (wish it was more plush but its ok enough) and of course the supercharger network is a 'killer app' for any tesla car.
autonomy might have been dangled in front of peoples' eyes, but anyone with any experience in software would have guessed that what elon keeps saying is just not in line with current reality. this car will NEVER drive itself across country. that's just a flat out lie and he should be taken to task for even suggesting such a thing. that's the crap I dont like about the company and elon. but the car is good for what it is (ignoring the direction the UI is going on, for the moment) and if it never gets above level 2, its still a great car.
set your expectations properly. not what some salesman says. and yes, elon is a salesman, first and foremost.
Actually I don't think Elon is a salesman at heart nor a very good salesman, but he's been convinced by others he has to be. His degree is in Physics, but at heart he's more of an engineer. He is aware that the media has a fascination with him and that is an excellent vector for selling Tesla to the world for free. He recently said he had Asperger's (a mild form of autism) and most people with it struggle with social situations and they frequently tell the wrong thing at the wrong time.
I think Elon is convinced that L5 FSD is possible. He's a brilliant designer and has some real skill as an engineer, but the self driving system is probably the most complex system he has ever attempted. He has written software, but it was for large systems that would be accessed with browsers. Embedded software is a different problem with different considerations. He underestimated what it was going to take.
Full self driving is a very difficult problem to crack. The problem of controlling a car on a limited access road like a highway in good weather is not that tough. AP1 is pretty good at that and it's vastly less sophisticated than the current FSD computer. But get onto surface streets with pedestrians, cars coming and going from driveways, bicycles, things coming into the street unexpectedly like a ball or someone playing with an RC car, it gets difficult. Add in weather problems and the problems get even worse.
To get to L4, the system can handle almost every case but there is still a human fallback for those cases the car can't figure out what to do. With my AP1 system there have been several times when AP was unavailable because the camera was blinded by the sun. They may have dealt with that with the newer system, I don't know, but there will be times the sensors can't get enough information to continue safely. Especially on a city street. There will be other edge cases where the car is confused and a human needs to take over. It may only happen a few times per hundred thousand miles, but with a fleet of millions of cars it's going to happen multiple times a day all over the world.
L5 has to make the leap from getting confused and having to have a human take over once in a while to never getting confused under any situation. From an engineering perspective, that's a very tough hurdle to overcome. For most engineering designs, getting to 90% is pretty easy, that last 10% can bankrupt you.
That's why the super high end gear for many things costs and arm and a leg. Back in the day when people had home stereo systems with separate components. There were lots of cheap systems for $100-$200, and a high quality system might run you $1000-$1500, but if you wanted the best audio reproduction possible, the amplifier alone cost $25,000. And this was back in the 80s. The $25,000 system was maybe 5-10% better than the $1500 system, but it cost more than 10 times more because it takes a lot of special engineering and construction to get that last 10%.
L5 has to be perfect 100% of the time. Tesla and other automakers might be able to achieve some kind of L4 system and maybe L5 on limited access roads in good weather, but the promises of all the radical changes to the world won't happen unless they achieve L5 and I think that's a bridge too far.
There are many business interests who are pushing hard for L5. It would eliminate truck drivers, Uber drivers, taxi drivers, and many other types of drivers. The economic benefits for businesses is huge, though the economic downside for people who drive for a living are also huge.
I could be wrong. But I think high reliability L4 is the best the auto industry is going to achieve. Some may try to sell their system as L5, but real world testing will show it can't handle every situation thrown at it. Maybe 99%, maybe even 99.9%, but not 100%.
