Tesla FSD 12 Needs Improvement
Vložit
- čas přidán 14. 06. 2024
- In today's video we see how well Tesla Full Self Driving Beta v12.3 can handle the infamous twisty Vermont Street and the Whole Foods Parking Lot
We post daily on Twitter: / aiaddict1
__________________________________________________________________
0:00 Introduction
0:53 Leaving Parking Garage (Attempt 1)
2:43 Parking Garage (Attempt 2)
3:47 Start of City Driving
4:10 Fails to Detect Proper Speedlimit
5:17 Scooter Edge Case
6:05 Vermont Street (Attempt 1)
7:05 Enters Private Driveway
7:40 Too Close For Comfort
8:04 Almost Clips Parked Car
8:19 Comparing v11 and v12
8:59 Vermont Street (Attempt 2)
9:30 FSD Gets Stuck
9:55 Confuses Fire Alarm Bell for Red Light
10:19 FSD ACCELERATES INTO CEMENT WALL
11:09 Too Close For Comfort Again
11:32 Almost Clips Parked Car Again
11:42 Conclusion
__________________________________________________________________
Tesla Referral Code: www.tesla.com/referral/john47584
__________________________________________________________________
The Best Tesla Car Fragrances (Link includes 15% Discount): www.dailylabshop.com/AIADDICT
__________________________________________________________________
AI Addict Merch: my-store-c00cd6.creator-sprin...
__________________________________________________________________
Business Contact: AiAddictOfficial@gmail.com
__________________________________________________________________
10% off Model 3 / Y Accessories; use code AIADDICT: www.vion.parts/?ref=27agak0846
__________________________________________________________________ - Auta a dopravní prostředky
Thanks for going back and doing the same road with v12. Looks like it's not ready for prime time yet but it's on the right path
Almost level 2.
It's still almost level two but you're still going to be stuck at level 1. I can't wait till I get it up to level 3 4 & 5, when will see what it can do there.
Yes, almost Level 2 after 4 years in beta testing.
With a huge team of engineers and programmers.
even if its 'almost level 2' its not even pass level 1.
@@deficator750 Level 1 or 2, just look at 50+ videos here on CZcams: v12.3 FSD still struggles a lot, and absolutely needs constant human supervision *and* *intervention* .
After almost 4 years in beta, FSD cannot even drive as well a high school kid who just got his license.
It's super fun if the only attraction you enjoy at Disney is "Toad's Wild Ride".
So true!
Thank you! You're one of the few I trust for these FSD reviews.
Awesome overhead angle
Thank you! First time trying it, I think we will continue to do it
FSD 12 is giving "white knuckle driving" a whole new meaning lol
Works flawlessly on real roads, it will expand to these torture test environments in time. Bias, dumb tester
This would be great software for my next Robovac.
👀
I've been rewatching the episodes recently, it's good to see you're producing new episodes. Other FSD channels aren't quite the same.
Great to see more perspectives. Thanks for the video 👌
Our pleasure! Glad you enjoyed it!
Really telling that it's still struggling. Looking closely, I suspect it had problems because it looks like there is a bug-splat right next to the camera that has been streaked over the camera. It's a good test because *thats what real car windshields look like*. If fsd can't handle any haze on the camera while driving in variegated light (through leaves) while the camera isn't perfectly clean then I have less and less faith that it can improve. (caveat: I want it to improve...I bought the damn thing)
It will drive straight through clearly visible road cones that have closed a road (and the signs for road closed was also clearly visible).
It does not have to do with bug splatters. It is just a really shitty software when it comes to self driving.
so nice that you are back making videos about teslas fsd! You are the only channel really testing the limits of this software!
Glad you like them! Hope to keep them coming :)
He is definitely not the only channel testing the limits of FSD. I have seen many videos of FSD being put through it's paces.
Omar got butthurt and blocked this guy. 😂
👀🤣
Back in Jan 2023 Omar posted s video of one of his late night drives. Dan O’Dowd spotted an intersection on a diagonal angle where Omar blasted through the red light. I think Omar took it down afterwards. I’d like to see if that or similar could be tested again to see how that works.
to be fair Omar is not the only one butthurt
@@AIAddictI'm assuming that before this summer arrives, your vehicle will have some sort of damage. Keep me updated! 😅
Omaris a tesla fanboy lmao he will never say anything bad about Tesla. I literally saw him almost crash and he'll say but it's getting better. Like bruh did you almost not lose your life and you're talking about it's getting better.
This was a fair review
In my experience v12 is vastly better than any version before it for the routes I take and doesn't do the abrupt breaking thing every time something moves around it.
Vastly better than a horrible v11 is not saying much. When can FSD be trusted with your life? When will Level 4 or 5 autonomy happen? Long way in the future.
@@DerekDavis213 Never.
Thank you and please more waymo tests
Will do!
do you think it'll hit that road cone again in sj?
Oh yeah…let’s go find out!
@@AIAddict 😄
Did it actually hit it or was that emergency breaking kicking in? (and if breaking was it the car or the driver?)
the driver manually hit the brakes
Driver manually brakes. That was NOT the AI emergency braking
can we have the top view camera angle for when the car accelerated into a wall ? it's strange it's not here, it would be informative to know how close it was.
Cool video anyway, thanks !
Noted! I can post on Twitter
Omg when it almost threw y'all into that short wall! 😳 Yes that's FSD alright: Fully Suicidal Driving. Elon should be ashamed of trying to pitch this as fully autonomous any time soon. I wouldn't trust FSD to drive down to the mailbox. Good video.
It’s a difference between stress testing the fsd and putting it in situations that it wasn’t designed to be in ( yet) . It’s almost like saying let’s see how fsd navigates under water! Newsflash FSD wasn’t designed to navigate under water so when it fails why are you surprised? I don’t believe the Parking garage was a fair assessment . I don’t believe it’s designed to be used in that scenario and using it in that environment is actually irresponsible OF THE DRIVER. Most people will activate fsd when they manually navigate out of a garage. You’re one of those people who give fsd and Tesla a bad name because you use it as it wasn’t intended to be used, and when something bad happens all the headlines says Tesla auto pilot crashes in parking garage. When it was not intended to be used in a garage. But enjoy your views buddy.
100% correct!
@@TeslaElonSpaceXFan indeed
Robotaxi, takes you wherever you want from wherever you are. Next year.
Every year. For ten years in a row...
Yes, he promised robotaxi next year, every year for ten years!!!
Please wake up. This is all a scam and will never happen with Tesla.
Did the car actually crash or did the car stop hard ?
I hit the brakes manually. Had I not intervened, it would have gone straight into that wall.
Crashed while accelerating. Driver hit the brakes.
Thanks for an honest review. Unfortunately a lot of scam from fanboys there.
it wasn't that honest. It goes out to paint V12 in a bad light. Which I have no problem with, but you have to show the 90% stuff it does well.
@@thewatcher5822 90 percent capable is NOT good enough. The whole idea is eventually Level 4 or 5 autonomy. That remains far in the future.
@@DerekDavis213 I agree 90% is not good enough. FSD is probably better than 90%, it was just a random number I threw out. But level 4/ 5 is a lot closer than people realise. It wouldn't surprise me to see it happen by the end of the year.
@@thewatcher5822 Right now, after 4 years in beta, FSD is a buggy Level 2 assistant. FSD still needs mandatory human supervision, every second on the road.
No way that will become Level 4, in the next 8 months!
I could list a dozen simple obvious things that FSD still cannot do.
@@DerekDavis213 Oh I am not saying FSD is ready. Even Tesla is saying that, and it will remain level 2 until Tesla say otherwise. But do not mistake FSD of four years ago with what they are doing with V12, it is a very different beast, moving away from heuristics to an end to end neural net.
Obviously I cannot say for certain it will be ready in a year. But from how V12 is performing, How capable ai is becoming and fast it is growing. The huge amount of data Tesla has built up, and no longer being compute constrained. I would certainly not be betting against level 4 by the end of this year.
Do you have a sunshade in your car? This is one of the few FSD videos where the drive is challenging. 12.3 seems to be about pumping the stock. Keep speaking Truth to the EV Power.
No, I do not own a sunshade. Glad you enjoyed the content! :)
Do you have some sort of beef with tesla?
No, I just enjoy testing and experiencing edge case scenarios and seeing what the software is and isn’t capable of
@@AIAddict sure !!! Why dont you tell the truth!!! Ha loser?! You are a tesla ex employee that got fired!!!
@@AIAddictlol you're a bad liar
Tbh you're going to be the laughing stock of the world soon. 🤣
At this rate of progress, we'll have SAE 3 when it's 1999 again lol.
By this do you mean never? Lol
@@AIAddict I think he takes into consideration the roll over bug at year 2999 when it becomes 1970 again.
Thanks for putting your money where Elon's mouth is. Causing an accident wouldn't be a fun experience, so thanks for taking the risk and sharing the results!
No problem, glad you enjoyed the content. It’s important to see the limitations and not become complacent :)
And that's what scares me. FSD, even if it does improve, will only get WORSE danger-wise as it does. Because then you become more and more complacent, and are less ready to respond when FSD messes up. I guess right now that's not an issue because it's still officially level 2, Tesla does not and never will take liability if FSD is engaged in an accident, and the system forces you to pay attention at all times. But it's something to worry about nevertheless. @@AIAddict
@@dahitmann Totally agree with your observation. The better it becomes and fewer mistakes made, the more dangerous it gets because drivers will become less vigilant. The problem will be even more deadly at highway speeds but even in slower city driving could be deadly for pedestrians.
@@dahitmann The is a fair point. The more comfortable a driver gets with the system, the less vigilant he will be, But being a level 2 system responsibility will always be on the driver. But the Pace Ai is progressing, It won't be long before Tesla deem it ready.
Musk said that same thing in 2014. It's 2024 now, 10 years on. Sorry if I'm skeptical it'll ever happen, much less _soon_ @@thewatcher5822
I wouldn't care too much about snake roads and parking garages at the moment. In the worst case, a robotaxi could leave that areas out. There's more important stuff to be fixed like construction stops and navigation issues.
What's nicer about v12.3 is the car blinkers are not going off on every curvy winding road. The wipers are also not going on as much as they use to. True some stuff v11 could handle were v12.3 now fails. It seems it's one step forward and 2 steps backwards. Will have to wait and see how fast Tesla iterates to improve as there no longer computing constrained.
Purposely chose the garage as a start. Personal agenda.
so it drives in circles...cool.
As you have asked, I want you to test it on Mt. Everest and under water please. There is barely any space to fit a car to drive that hill, even a normal human being will be scared to drive there, don't see any road marking either. Can't blame tesla for an underground garage exit, they have bigger things to fix with the FSD. Hope you are sending these data to tesla so they can train it for future updates.
From what we know. Tesla has not even trained on parking lots yet. Emphasizing this is not pointless until they actually train for it.
I do make this exact distinction within the first 2 minutes inside the garage. It was still a fun test regardless. I cannot judge FSD for how it handled the parking lot, but as for the rest, it has no excuse for being so bad.
It was bad on only one street. That is a very complicated street that Tesla is not even training for. Overall, the point is that you are having to test rare situations to show its limitations. The stuff you tested two years ago, V12 is really good.
@@MrRaowens1 He does seem to be going out of his way to find challenges for the system, which is fine by me, as watching V12 navigate through streets can get a little boring, as it is that good. What I would like to have seen is him doing the test 4 or 5 times, to see if the system can figure it out. I have seen this in other videos.
@@thewatcher5822He was fired from Tesla. He is still bitter.
@@MrRaowens1 Yes there is certainly an anti Tesla bias to his videos, So I take the finding with a pinch of salt. But by the same token he is digging out edge cases where it still needs work. I'm just interested to see how it performs. On the whole V12 is proving its worth. But it is not ready, and these sort of videos show how close or far it is to being ready.
From what I am seeing FSD is very close.
Watching this video at length, it really begs the question. What is the point of this FSD system, exactly? Is driving around on residential streets and in parking garages the issue, in terms of the problems FSD is intended to solve?
To me, it looks like they're wasting 99.9% of their time and resources trying to perfect something that constitutes only a tiny amount of what makes driving boring drudgery. I think for the time being, FSD systems should cede urban areas and residential zones to manual control, and just make sure they can do freeway traffic perfectly. I owned a Model S 100D between August 2017 and early January 2024, and the only time I really wanted to use FSD was on freeways in Seattle. It was just so much easier driving manually that last half mile or so from the freeway to my house.
This current iteration of the tech is a solution in search of a problem, and it's not likely to ever be better than being driven around town by your 90 year grandmother with bad cataracts.
Most car companies already have a good solution for freeways, and it is indeed where you actually need it. The problem is if Tesla doesn't have the FSD that Elon promised then they are nothing without it.
Brian wrote "FSD systems should....just make sure they can do freeway traffic perfectly"
Yes but the latest FSD has problems with that too: cannot drive at the proper speed limit, problems with Left Merge for example, and not getting in the proper turning lanes, in time. lane changes in bumper-to-bumper traffic. The list goes on.
@@DerekDavis213 Thank you for the info! When I talked about my own "FSD" use, I actually just meant the standard level 2 autopilot. That system had some fatal flaws in it for sure, but seems like it was much more reliable than this new thing.
The point?
Pump the stocks of course. Especially now. Elon desperately needs the money.
@@Xanthopteryx Elon needs money for his Mars colony.
Elan said 1 million humans on Mars by 2050. That is just 26 years away. Ridiculous!
thanks for sharing the real status and progress of V12
Real progress. V12 is a huge leap over earlier versions. It is not ready, and I am sure there are many places where FSD will struggle. Parking garages an obvious one. But overall there is plenty to be excited about when it comes to FSD v12.
@@thewatcher5822 Yes it's definitely better but took way too long to get to this point, probably at least 10 years before L4 sadly.
@@Starshipsmm To be honest, I think a year from now, Tesla will have a very reliable system. They have all the data they need, they are no longer compute constrained. There is a lot to be positive about. Time will tell, but I think Tesla has found the solution to self driving.
How did they go from compute restrained to not compute within a month? Something about Elons tweets don’t add up.
@@AIAddict They have probably bought more compute online. It isn't just a matter of clicking your fingers. It takes time to put in place. I do know they have been ramping up Dojo, so that is probably a big factor.
As far as full self-driving for Tesla i can already tell you right now full self-driving does not work in underground garages, it's like a cell phone signal it fades in and out. But that's why it needs some work!
I think a lot of humans panic when going through that road.
Still some learning required though.
oh man i should had got a bucket of pop corn for that one!.... you 2 are brave souls!!! 💪🫡
🤣 glad you enjoyed it!
Way too close to the walls for *your* comfort. I routinely drive within 3" of obstacles at slow speeds also. No need for cameras or sensors, just know your view angles and reference points and it's fine.
You routinely drive to within 3 inches?
Well there is a huge difference between you and the beta FSD software: human brain and human stereoscopic vision!
total the car at 4mph? 😭
The world may never know 🤐
Car camera placement not so ideal. Fsd future looking promising.
Agreed, Tesla needs to add cameras to front bumper or add new sensors.
It shouldn’t get that close to parked cars. Someone may try to get out. Scary it almost sent you into a short wall. Very scary!!
With no ultrasonic sensors, or Lidar, or Radar, the car cannot measure distance. A big problem.
Watching while waiting for Tesla service. Just curbed and popped the tire :/
Lots of anger in this video.
☹️
And now 12.3.5 🙂
Robo-Taxi next year ... said the Tesla bulls.
Supposedly we will have an unveiling in August 👀
@@AIAddict Press X for doubt
Love the fsd but don’t be blindsided and call something perfect if it still needs a bit of work which is fine . I do own a model s if u think I’m tesla hater I’m not but this video says things for what it is rather than that Omar guy videos which always shows only good thing . Try the fsd in New Jersey I can’t wait to do that
Well said. FSD performs some tasks amazingly but you cannot allow yourselves to become compact because it can and will mess up. I feel I personal fall in the middle
You're the opposite of Omar Qazi. He does Tesla propaganda by only showing cherry-picked good drives, and you like to challenge it and show everything.
I learn more from your drives; keep it up!
Well said, thank you for your comment! Welcome to the Channel :)
Chuck cook was impressed with the latest version as well. But they seem to be training the car in Jacksonville now.
How on earth would Omar cherry pick good drives when he puts out hours of unedited videos non stop?
@@jiggig By Omar not showing his accelerator presses, by driving in easy conditions and routes, and not showing the bad drives.
A real AV tester in San Fran, such as Waymo test drivers, would relentlessly drive in areas where they have problems, seeking out the problems to overcome them.
To get a driverless license in a real city like S.F., the AV needs to show millions of safe miles without exception. Not a few short cherry-picked "zero intervention!" drives.
Science progress is done by scientists who spend their time trying to disprove their own hypotheses, the essence of falsifiability. Find one bad result and the hypothesis is wrong and thus needs more tweeks. Propaganda is the opposite of science.
very different with other FSD12 viedo
Feeding the Tesla haters one video at a time. 😂
😂
Its not even enabled yet in parking lots
I do make this exact comment within the first 2 minutes inside the garage. It was still a fun test regardless. I cannot judge FSD for how it handled the parking lot, but as for the rest, it has no excuse for being so bad.
@@AIAddict Was it so bad? It didn't know the way out, with lots of cars around. It is basically what I would expect.
@@thewatcher5822yes, crashing into a wall is bad
@@WatchNoah I was talking about the parking garage. Crashing into a wall is a definite no no.
should rename it full distracted driving, having to constantly baby sit it definitely takes your attention away from actually safely operating the vehicle.
It certainly does, I like that name you came up with. It’s like being a driver instructor. Sometimes relaxing, other times anxiety driven.
4:50 Well, you did employ yourself as a beta tester :P
7:30 It's not interesting, it just reveals a basic lack of quality control at Tesla, like regression testing.
Yeah, Quality Control and Tesla are not often in the same sentence.
Funny seeing how these Tesla apologists were blasting you for not showing v12 in your last videos because it's so much "better" and now you have this... These fanboys need to be more realistic about what's going on here. I personally think people need to take a hard look at what Google is doing with Waymo and see how it's not even close.
Exactly! They screamed into the void “but v12 will fix this!” but in reality v12 performed sooo much worse than v11.
It’s clearly evident that Waymo is the superior robotaxi king.
@@AIAddict I personally will only consider a Tesla FSD if they put back lidar and/or radar. They dropped this sensor fusion because of Elon's cost-cutting philosophy and you're seeing it play out in comparisons. This video is shot in perfect conditions in San Francisco. I don't want to know what it's like in a sub-optimal environment with this camera-only solution.
@@AIAddict They are two very different systems. And while Waymo is certainly the safer option (it has to be it is level 4) Tesla is taking on a much bigger challenge. If they can work it out, they will be in a very dominant position. V12 is a huge step in that direction.
@kristjanwashere Waymo is not even the same thing. It doesn't scale. Tesla might never get it to work, who knows, but they are trying to make the car drive without every street having to have a highly detailed 3D (prerecorded) mapping, like Waymo does. Imagine the effort it would take to do this for every road and keep them up to date. They can't even keep low detailed maps up to date.
@@AIAddict how? Waymoe wouldnt even take this kind of road.. i wonder if you can force it to take this road? just curious
Hey everyone. he is a tesla ex employee that got fired!!! Just you know !
It was my job to validate FSD before it went into customer hands. If anything, my opinion should hold stronger.
more video😂
We have more planned!
If I would have a Tesla and did pay for the FSD I would be very pissed. This is so but SO far from being safe and complete.
It is far closer than this video might show. He is going out of his way to try and trip the system up. It is not ready for a garage, and a few more attempts down that road would probably seen the car figure it out. Updates are coming a lot faster now. By the end of this year this could well be a level 4/5 system
I can guarantee you FSD will not be Level 4 this year with current hardware. I’d put money on that.
@@AIAddict Follwing self driving as I do, It is not a bet I would be willing to make.
@@AIAddict That is certainly not a bet I would be making.
@@thewatcher5822 Absolute delusion.
cut it some slack. I understand your frustrations but I think what is important is to notice the pace of improvements. I think you'd agree v12 is MUCH better than v11, even if you might still not like v12 very much. The direction things are going in is more important than their current state
I much prefer v11.4.9 over v12.3
@@AIAddict Sorry. From what I am seeing V12 is far superior. Sure it has taken a few steps back in some areas, but has taken a huge leap forward in others.
He doesn't show his FSD settings, so either has no idea how it works or ggosh has an agenda.
I'm actually super impressed that this is their first version of end-to-end, and it's already this good. You have to push the limit to break it. I can't wait for version 13 or 14, especially when they train it for parking lots and auto park functions. Keep pushing the limit so they can collect the data.
Absolutely. I have been watching many videos of V12, and to be honest, it is getting that good they are getting boring to watch. These more extreme cases are where the car and Tesla will learn the most.
I would like to have seen 1 or 2 more attempts, as I have seen V12 learning very quickly how to do things. Surprised to see it heading for a wall, but this is why its still in beta.
We wanted to try a third attempt, but thought it best to quit while ahead before an even worse incident possibly occurred.
Elon said FSD 12 wasn’t beta
@@videos-th1ivhe says a lot of stuff that is not true today.
@@videos-th1iv Well he can call it what he likes. Until it runs without humans, it will always be level 2.
V12 does not learn to do something. It should do the same thing every time. If it doesn't then the system is not working correctly
looool this is barely L2, FSD my ass
It is level 2 until Tesla say different. But it is far more capable than that.
@@thewatcher5822 No it's not.
@@Xanthopteryx Well obviously it is, as it is doing most drives now with no human intervention. Moving away from heuristics has proved an inspired move, and the car now drives in a much less robotic way and feels far more human.
@@thewatcher5822 It is not more capable than an ADAS system. Not at all. Not even close. It can not even handle basic simple tasks like... stay on the road, do not drive into stationary objects, do not run red light, and so on.
@@Xanthopteryx It does all those things, it makes the odd mistake, but most of the time it is driving perfectly. Most importantly it is doing this across America.
The fact that you still double-pull the stalk to engage FSD tells me all I need to know :)
Classically conditioned to double pull, been using FSD for 3+ years and this is the first single pull version.
Can I have some of your kool-aid?
@@Mark-kt5mhno
I can't help but feel for the tesla, I'd be suicidal the second time down there...
As a human, it’s actually not that bad. I’m able to easily average 15mph with one hand with no issues.
까고 싶어서 만든 느낌이네. 경로도 지정 안하고...안 내려간다 하고..
What a shitshow
😂
My experience with 12.3 is night and day, closer to Omar than you and I have no favorites in this fight
If you dont see that, you are an outlier
As I say in the video, FSD 12 handles some situations with perfection. It certainly behaves different than V11. However, other situations have regressed as seen.
Let’s see Omar livestream the same drive. Of course Tesla will have to HD map the drive first to give the edge.
Waiting for CHP to pull Omar over on one of his long drives for unsafely operating his vehicle (if he is using a safety bypass device on the steering wheel).
@@allenbaylus3378 lol , my wife is not complaining anymore with 12.3 last 4 days. I could
Not use 11.4.9 with her in car before.
So it’s way better, and that is why Elon is forcing demos as of tomorrow.
Many people bad mouthing FSD progress are doing it due to politics I feel. 😀
wank.gif
@@shdmd2118 Yes, videos like these mask the real potential of V12. These are rare cases, the fact it starts in a parking garage shows that. I'm pretty sure if he had done that run down the street a couple more times, FSD would have figured it out.
What an atrocious pile of garbage. Imagine how many people it would kill if it was left to it's own devices. Hard to believe it's legal to be honest.
It seems pretty irresponsible of you to use this feature in public.
Tbh it seems pretty irresponsible the NHTSA allows Tesla to ship this to 400,000 people
No improvement in past 5 years lol
Pretty much :/
Actually massive improvement, the disengagements are far fewer, and the driving is becoming almost human like, handling some really complex situations that would have tripped up earlier versions. V12 could finally be the key that unlocks self driving.
Welcome to the 🤡 channel!
What a clown video 😂😂😂
Butthurt ex employee does test drive in parking lot. FTFY 😂
These are stupid exaggerated difficult tests for where FSD AI is right now.
They are not stupid. They do exist. He could give us an hour of it driving around, but that would be pretty boring. He is challenging the system, and that will only be good for the FSD and Tesla. Data is everything. I just wish he would keep going with his tests.
What is particularly annoying about this guy, and there are others that are like this, is that they talk as though the current version is the final version of the software, as though Tesla isn't pushing updates every couple months, and in some cases every couple weeks. There's also this tone to the commentary that implies that Tesla engineers aren't aware of the bugs. This guy is a clown.
If Tesla engineers were aware of the bugs, then why would they let half baked code leave the shop?
That would be criminal.
@@AIAddict You just have poor comprehension of the complexity of the problem being solved and Tesla's strategy when it comes to solving it.
Waymo has a great product right now, but a Waymo won't drive down that Vermont St. at all, they geofence it so that it avoids really difficult edge cases. Waymo AV's also don't drive into or out of parking garages. So Waymo has solved the "curated and micromanaged AV" problem (they have support staff that intervenes remotely and very frequently).
Tesla is trying tot produce a level of autonomy that works in a much, MUCH broader range of scenarios and with minimal remote monitoring and intervention. If they went about things the Waymo way it would be a non-starter.
If Tesla is successful it will save millions of people from the burden of car ownership and it will eventually save tens of thousands of lives each year. They should be applauded for their efforts.
You should run a channel that reviews the latest gadgets and whatnot, because that's the mentality you are bringing to this. You should be reviewing the latest Roomba.
@@AIAddict It would be criminal if Tesla sold it as a level 4/5 system. They have always said it is a level 2 system requiring the drivers attention at all times. There are kinks to sort out, But V12 is looking the real deal.
What's even more annoying are the deluded Tesla brown nosers like yourself who are completely detached from reality.
@@paulhamrick3943 I agree the @AIAddict has a more negative stance on FSD (I watch a lot of positive of FSD videos and I'm very pro FSD but he a counterbalance for me) but you are wrong about Waymo and Vermont Street. He posted a video a few days ago about a comparison with FSD11 and Waymo doing the same street and Waymo (geo fenced most places not here now) did an amazing job. this is sort of a follow up for FSD12.3. Looking forward to FSD 12.3.1 (12.3.2 released very soon) and above videos for this.
I don’t know what point you’re trying to prove here ? Fsd works great for me maybe you need a service apointment
That’s the entire point. It works and sometimes it doesn’t. It’s not done yet, as you know, since you read and clicked the agreement. That’s not a bad thing. It’s a thing that *everyone* is aware of. Great progress is being made. But, it’s not done yet, obviously. Tesla is perceptive of this reality, which is why it’s still beta, and you have to keep your hands on the wheel.
@float32 gets it ☝🏼
the amount of videos of fsd failing and the sheet amount tesla fan boys getting butthurt and making excuses for the fsd is so pathetic. I own a tesla but i have given up on fsd and autopilot its extremely dangerous.
From what I can see, you have a spite with Tesla and you just want to show the parts where Tesla has not have enough data to train their system yet. Especially parking lot and the winding street you are testing
It's been near 10 years now and the thing still cant handle a parking lot. Kinda makes me wonder why after all this time you still believe Elon Musk after all these years when there were suppose to 1 million Tesla Robotaxis on the road by 2020. What was he seeing back in 2018 to possibly think they'd be ready by 2020 when here in 2024 it still looks like a terrible experience .
@@justin9152 it can handle turns, freeways, city streets. But not yet the parking lot and extreme winding street like the one AI ADDICT was testing
Look at our last video, it handled Lombard St very well which is just as curvy
Yeah but still I wonder why you focused on this edge cases?
Winding roads are not an edge case, they are almost the norm in other counties
We both know it is way too difficult to AI at this moment, and this road and parking garage are very rare user case in daily life...
As long as FSD / AP can resolve 90% of daily user cases, I am happy.
Well in that case for you, FSD is just an ADAS system. But Tesla wants to sell it as a L5 system so it needs to be able to handle parking garages and this road 1000x without incident
@@speedismyfriend I am not a big fan of Tesla, and I don’t care Tesla wants to sell it for L2 or L5. That part is just commercial stuff. Not important at all.
I care more about FSD and AI itself.
It makes much more progress than I expected, handling 90% of user cases.
FSD is extremely important as an indicator for AI applications. We now can see some hope that AI actually changes life.