Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Why is the auto parking so bad? 8 Years later.

This site may earn commission on affiliate links.
FWIW I've always had really good luck with autopark. I've used it 100's of times, and only had to abort parallel parking attempts twice now. I say "good luck" because apparently I'm in the minority.

Edit: it occurred to me that I have barely, if ever used autopark since getting the HW3 upgrade, due to the pandemic and being basically unemployed. So if something has changed, as knightshade alludes, I'm not aware of it. Yet.
 
FWIW I've always had really good luck with autopark. I've used it 100's of times, and only had to abort parallel parking attempts twice now. I say "good luck" because apparently I'm in the minority.

Edit: it occurred to me that I have barely, if ever used autopark since getting the HW3 upgrade, due to the pandemic and being basically unemployed. So if something has changed, as knightshade alludes, I'm not aware of it. Yet.


the hardware has for sure changed- it's possible they're still running the same code though, just emulating it on the much more powerful hardware instead of running it naively on a lesser chip that no longer exists- at least until they've got better (reverse summon?) native code to use.

Seems like a good question for someone on the twitters to ask Greentheonly
 
  • Like
Reactions: sduck
the hardware has for sure changed- it's possible they're still running the same code though, just emulating it on the much more powerful hardware instead of running it naively on a lesser chip that no longer exists- at least until they've got better (reverse summon?) native code to use.

Seems like a good question for someone on the twitters to ask Greentheonly
It’s a good question.
I hope they are using the same code. Imo that would suggest it’s just a bridge until they’re ready to implement the equivalent aspects of FSD. Which in theory could be somewhat soon, even if regulatory and technological challenges delay FSD to a much later date.

I don’t need FSD, but I’d pay for 8k for things like better blindspot and cross traffic warnings, backup positioning etc. Things I actually would have used, (and would use in the next few years). I wonder if Tesla is hardware limited at this point though. Those options get expensive to produce. The allure of FSD is much more digitally based, with higher margins.
 
  • Like
Reactions: Kweber!
I remember my dad showing me the automatic parallel parking on his Mercedes five or six years ago. It did it on the first try (reversing into the spot, then moving slightly forward to center itself in the spot). I was really impressed. So yeah, I am sort of disappointed that Tesla doesn't have this figured out (I don't even have FSD, but I've seen how less-than-stellar it is from videos).

My personal thought is that when Tesla gets parking figured out they should make it part of standard Autopilot (not just FSD). I mean heck, the Toyota Prius which starts at $25k has it as a standard feature. The same goes for automatic lane changes, should be part of Autopilot, not FSD. Lots of cars these days come with lane-keep assist/adaptive cruise control (any car with Honda sensing does it -- my wife's Honda does a good job at staying in the center of the lane and going around curves on the highway). Yeah, I know this will piss off lots of people who paid for FSD, but I think people associate Teslas to be better at this stuff out of the box. FSD can be reserved for anything off highway, or navigate on autopilot, etc. In a number of years, Tesla will probably have to include the entire FSD suite of features as standard anyway as other manufacturers catch up.

Well this is true a lot of cars seem to do this fine and faster.

You realize nearly every legacy manufacturer had better years ago?

I am afraid this is also true.

It is interesting that the supposed leader in FSD has one of the worst auto parks. Perhaps a function of not having the resources available to work on something that will ultimately be replaced by the end-product FSD?

On the same note, Tesla has inferior backup and blind spot HMI to similarly priced competition. This also makes some sense if the end goal is FSD and you’re resource constrained, but in the meantime it’s one of the many little things that give the competition a chance.

Its actually a paid for feature of FSD that is meant to be implemented fully but obviously not as refined as the other manufacturers who give it away free with the car. Blind spot detection is very hit / miss for me when tested before changing a lane. Technically other cars lane keep without doing strange stuff like the FSD autopilot crossover.

For all the FSD believers, this thread is a good example of why your existing Tesla will never be a RoboTaxi for you or anyone else. Tesla is only a leader in FSD in their own marketing material. They can't even get things like Autopark and Lane-keep cruise control to work right with their setup after many years working on it. FSD is never coming boys. Maybe some future generation of hardware will do it, but not what we own.

Well apparently it will be end of 2020....
 
  • Funny
Reactions: TLLMRRJ
It is weird. We also have a 2013 Ford Escape with auto park and it nails it every time! And does it very fast. The Tesla is a brain-damaged slug in comparison...

I really wish someone came out from Tesla with an honest answer. Its something that it should excel at with all those sensors. It must be able too unless the sensors are not precise enough. I know from parking in my garage the sensor bubble is very slow to react to its surroundings like it tried to interpret what is around it slower. Its like get the basics right there is a reason why people are not using it... are all the people working in FSD only?
 
  • Like
Reactions: BlindPass
I really wish someone came out from Tesla with an honest answer. Its something that it should excel at with all those sensors. It must be able too unless the sensors are not precise enough.

The only sensors the current parking code uses are the ultrasonics.

It's why you can only perp. park between 2 cars- it has no idea where the lines are (or if there even ARE lines) it just knows where the two big objects the short-range ultrasound is bouncing off of are.


The reverse-summon parking will finally use cameras and then it's likely to not suck at all.
 
IIRC, autopark is provided by a 3rd-party chip (Infineon Aurix). It will probably always suck until something gets properly baked into AP. Then it will suck less.

that does not make sense. the aurix (I am familiar with it, assuming you mean tricore chips?) is just general purpose compute and controller. unless someone else wrote firmware to autopark (doubt it), then your statement does not make sense.
 
that does not make sense. the aurix (I am familiar with it, assuming you mean tricore chips?) is just general purpose compute and controller. unless someone else wrote firmware to autopark (doubt it), then your statement does not make sense.


It's been known for years the aurix chip is used for parking and a few other things. See among other sources-

https://twitter.com/greentheonly/status/1210437926515023873

Green said:
It's also responsible for autopark and summon (explains why cameras are unused in those modes!)

Also the camera heaters

And yes- it absolutely is running its own custom firmware.

https://keenlab.tencent.com/en/whitepapers/Experimental_Security_Research_of_Tesla_Autopilot.pdf

this has some more detailed tech info on what the Nvidia vs Aurix chips are doing

Both APE and APE-B are Tegra chips, same as Nvidia's PX2. LB (lizard brain), is an Infineon Aurix chip. Besides, there is a Parker GPU (GP106) from Nvidia connected to APE. Software image running on APE and APE-B are basically the same, while LB has its own firmware.


HW3 lacks the aurix chip so obviously they've got either emulated code running or some native stuff written by now
 
Last edited:
For all the FSD believers, this thread is a good example of why your existing Tesla will never be a RoboTaxi for you or anyone else.
Taxis like Uber or Lift never park, they stay in middle of the street and block all the traffic while waiting for a customer.

And to even annoy more everyone, they stay at least three feet away from parked cars so customers will not hit a car when opening the door.
 
Taxis like Uber or Lift never park, they stay in middle of the street and block all the traffic while waiting for a customer.

And to even annoy more everyone, they stay at least three feet away from parked cars so customers will not hit a car when opening the door.
Right, this feature is much more appealing to the owner-driver model. If your long term view of ownership is more on robo taxis, mastery of this type of parking is lower in hierarchy.

Still, one would think though that a technology company like Tesla, capable of developing largely in-house FSD, would still at least have a bridge as good as the industry standard. Perhaps they view that having a quality interim feature as such a fool’s errand that they simply don’t care enough to compete.
 
Right, this feature is much more appealing to the owner-driver model. If your long term view of ownership is more on robo taxis, mastery of this type of parking is lower in hierarchy.

Still, one would think though that a technology company like Tesla, capable of developing largely in-house FSD, would still at least have a bridge as good as the industry standard. Perhaps they view that having a quality interim feature as such a fool’s errand that they simply don’t care enough to compete.

Yes bang on! Interim even we part our cars all the time yet one of the features like parking and even summon is dire near useless. Where on a playing fields on of other cars is something of a given for them. I just think its a practical feature that has been left to the side.

I've only ever been able to get to even think there is spot about 3x in 2 years much less park itself (never).

So no improvement in 2 years in this area?

I admit I am a Tesla fanboy but when I feel something should be better or done better I have no problem in voicing it here as I am sure they are reading our comments somewhere.

I have a theory is it possible by cracking FSD totally that it would then filter down into a much smarter summon and parking experience? If this is the view at the drop of smaller improvements then I think this is a mistake. Because in the chase for the holy FSD grail you have just lost all ground to the competition. Investment in the software should be done at both ends. Normal functions of the car and production should be better than the other manufacturers. I bought a model 3 in 2019 because I thought that was when FSD would drive on city streets.

You know this very evening I just parked up while my partner went into the store to grab a couple of things and these 2 kids go past with their parents and he goes “ OMG its a Tesla that is amazing do you see that “ not the first time I have seen youngsters totally inspired by the brand, the tech. Now would they if they knew that other cars can already do the features touted years ago better? Making FSD more expensive for the very people obviously testing it for you, saving their cars from driving off the road automatically for you to almost getting rear ended with phantom breaking. This is the fools errand I think as you are now getting less working features for your money.
 
Yes bang on! Interim even we part our cars all the time yet one of the features like parking and even summon is dire near useless. Where on a playing fields on of other cars is something of a given for them. I just think its a practical feature that has been left to the side.



So no improvement in 2 years in this area?

I admit I am a Tesla fanboy but when I feel something should be better or done better I have no problem in voicing it here as I am sure they are reading our comments somewhere.

I have a theory is it possible by cracking FSD totally that it would then filter down into a much smarter summon and parking experience? If this is the view at the drop of smaller improvements then I think this is a mistake. Because in the chase for the holy FSD grail you have just lost all ground to the competition. Investment in the software should be done at both ends. Normal functions of the car and production should be better than the other manufacturers. I bought a model 3 in 2019 because I thought that was when FSD would drive on city streets.

You know this very evening I just parked up while my partner went into the store to grab a couple of things and these 2 kids go past with their parents and he goes “ OMG its a Tesla that is amazing do you see that “ not the first time I have seen youngsters totally inspired by the brand, the tech. Now would they if they knew that other cars can already do the features touted years ago better? Making FSD more expensive for the very people obviously testing it for you, saving their cars from driving off the road automatically for you to almost getting rear ended with phantom breaking. This is the fools errand I think as you are now getting less working features for your money.
While I do think Tesla’s autopilot/pre-FSD is better than most at “driving”, the gap has narrowed and they’ve been surpassed in other tech. Now, if they release FSD soon, maybe they’ll leap forward again.

I give Tesla a lot of credit, that they’re even able to compete with such mammoth companies is impressive. But in many ways you can tell they’re caught in the dichotomy of pursuing a downmarket disruption (EV for the masses and robo taxis) with an expensive nascent technology production costs. To have an affordable EV that competes with ICE, they have to cut costs imo