GPS-based Parked Position Incorrect.

  • If you haven't taken delivery yet or plan on ordering you can still get the 6 months of FREE Supercharging only until December 31st all Model 3s now qualify! Call or email your Tesla delivery advisor and give them our code
  • Please post everything related to v9 (2018.39.x) into the version 9 MEGATHREAD. Please do your part to keep this site organized for everyone by looking around or doing a search BEFORE posting. Thanks

  • Winter is here and the forum is starting to get flooded with cold weather threads and posts. Please heed the suggested threads and posts before you post something related to cold weather. This is a great place to start: https://model3ownersclub.com/threads/teslas-in-cold-weather.5271
  • It's OK to discuss software issues here but please report bugs to Tesla directly at servicehelpna@teslamotors.com if you want things fixed.

Karl Sun

Active Member
Joined
Sep 18, 2018
Messages
194
Location
Arizona, Earth
Tesla Owner
Model 3
Country
Country
#1
When my Mdl3 is in my garage, the displayed [physical[ address is correct.

Move it 20 feet into the driveway and it displays the address across the street. Evens on one side, odds on the other. So the zero-point for the GPS location system to off by 40-50 feet.

Anyone have any idea how to fix this?
 

MelindaV

☰ > 3
Moderator
Joined
Apr 2, 2016
Messages
6,811
Location
Vancouver, WA
Tesla Owner
Model 3
Country
Country
#4
When my Mdl3 is in my garage, the displayed [physical[ address is correct.

Move it 20 feet into the driveway and it displays the address across the street. Evens on one side, odds on the other. So the zero-point for the GPS location system to off by 40-50 feet.

Anyone have any idea how to fix this?
others have made similar observations and found out it is an issue with the google maps data that Tesla uses. so what they use to pinpoint addresses on your street is probably off center a little. i've found the places I park to be very accurate.
 

Ed Woodrick

Well-Known Member
Joined
May 25, 2018
Messages
277
Location
Atlanta, GA
Tesla Owner
Model 3
Country
Country
#5
If it's gonna show where it's ,parked in the app, then Yes. It matters. Non-functional "features" are Bugs by definition.

BTW really cheap ($15) GPS units are accurate and repeatable to 9 feet.
Sorry, I believe that while your statement is correct, your interpretation is off a little. Yes, they can be that accurate, no, they aren't always that accurate. https://www.gps.gov/systems/gps/performance/accuracy/

I have seen GPSs with popcorn trails turned on, that have wandered thousands of feet over a day or two. There have been many people talking about how the Tesla has them on the wrong, but often parallel road.
And along this thread, addresses are generally listed as a range, along a road segment, with the even and odd sides marked. If houses are not even distributed, then address location won't be perfect. If the GPS is off by only 20 feet, it could put you on the wrong side of the road.

Most automotive GPSs implement an algorithm that "pins" a car to the nearest road. This falsely helps you to believe that it is really accurate. And this is where they sometime "pin" you to the wrong road.

If you were to use hiking or aviation GPSs, that don't pin you to roads, you start to see the issues a little better. They can get you really close, but they aren't perfect.

I remember seeing my first GPS in 1982. It was awesome! If you left it in the same place for about 2 days, it could tell you where you were to within a few hundred feet. This was so much more accurate than any other solution, aside from having a survey team figure it out.

Even with survey teams today, the use GPS, but they always have it referenced against a known location. They generally use 2 or more GPSs, and correct the second one based upon where the first one says it is while at a stationary location.
 

Karl Sun

Active Member
Joined
Sep 18, 2018
Messages
194
Location
Arizona, Earth
Tesla Owner
Model 3
Country
Country
#6
We have several GPS-based NTP Stratum 1 clocks at work (so we know the time to within sub-microseconds). The antennaes cost ~$75/US or so. They will give a location to within 0.1 feet, aka 1.2" with latitude, longitude, and altitude .I realize the electronics in motorcars isn't and doesn't need to be to that level.

Would be nice if I can "move" the GPS "center" location in the car 20 feet or so so it gives the correct address.
 

Lovesword

Top-Contributor
M3OC Supporting Member
Joined
Sep 12, 2017
Messages
1,229
Location
Iowa
Tesla Owner
Model 3
Country
Country
#7
We have several GPS-based NTP Stratum 1 clocks at work (so we know the time to within sub-microseconds). The antennaes cost ~$75/US or so. They will give a location to within 0.1 feet, aka 1.2" with latitude, longitude, and altitude .I realize the electronics in motorcars isn't and doesn't need to be to that level.

Would be nice if I can "move" the GPS "center" location in the car 20 feet or so so it gives the correct address.
I gotta agree, I hate that it shows my vehicle is parked over in the neighbor that I dislike’s kitchen.
 

GDN

Moderator
Moderator
M3OC Supporting Member
Joined
Oct 30, 2017
Messages
1,991
Location
Far North Dallas, TX
Tesla Owner
Model 3
Country
Country
#8
Guess I've got some good street data for my neighborhood, my car is perfect in the garage every time I've ever checked. Even where my partner parks at work I can look it up and know exactly which parking spot he got at work that day..
 

evannole

Active Member
Joined
Jun 17, 2018
Messages
55
Location
Marietta, GA
Tesla Owner
Model 3
Country
Country
#9
Mine is uncannily accurate, especially considering that our garage is in the basement and our house is four stories tall, including the partial daylight basement. The app accurately shows even what side of the garage the car is in. My phone usually has trouble picking up a GPS signal in the garage, but the car does not.

A few weeks ago, while I was driving, the GPS did misidentify our positon by about a half-mile. At one point, it thought we were on the runway at Dobbins AFB when instead we were on a state highway. The GPS lady was very confused; I half expected her to warn me of a C-130 on final. I rebooted the screen several times and it finally figured out where we really were about 10 minutes later, on I-75. I haven't seen this problem since.
 

stuff

Active Member
Joined
Sep 19, 2017
Messages
52
Location
USA
Country
Country
#10
When my Mdl3 is in my garage, the displayed [physical[ address is correct.

Move it 20 feet into the driveway and it displays the address across the street. Evens on one side, odds on the other. So the zero-point for the GPS location system to off by 40-50 feet.

Anyone have any idea how to fix this?
At least one of use (perhaps both) are outliers from the norm. My GPS location has been ridiculously accurate in the 1200 miles I've put on the car so far. So far I've never seen the GPS location off by even a parking space from it's actual location. When using the app on my iPhone and it shows the blue dot for the phone's location, the phone's GPS is total cr@p by comparison.

I am a pilot and fly with DGPS, to be honest the Model 3 "feels" as accurate as my DGPS units - I'm sure it is not, but you couldn't tell it by my observations. Prior to your thread I have been asking myself what tricks they are using to make the system so accurate as I've never seen normal GPS units as stably accurate as the Model 3 is performing. Best guess is that they are augmenting with inertial processing tricks.

As Melissa points out, your address resolution question is a different animal - with errors coming from other vectors. That's not a GPS issue.
 
Last edited:

Ed Woodrick

Well-Known Member
Joined
May 25, 2018
Messages
277
Location
Atlanta, GA
Tesla Owner
Model 3
Country
Country
#11
I am a pilot and fly with DGPS, to be honest the Model 3 feels as accurate as my DGPS units. I'm sure it is not, but you couldn't tell it by me. Prior to your thread I have been asking myself what tricks they are using to make the system so accurate.

As Melissa points out, address resolution is a different animal - with errors from other vectors. That's not a GPS issue.
DGPS is a very different animal. With GPS, when off by a little, generally everything in that area is off by the same amount. DGPS is the solution to the problem, because it places a GPS at a known point and then broadcasts it's correction figures to other locations. You then get the correction to apply to what you are receiving and your accuracy can get very accurate.
 

Ed Woodrick

Well-Known Member
Joined
May 25, 2018
Messages
277
Location
Atlanta, GA
Tesla Owner
Model 3
Country
Country
#12
Mine is uncannily accurate, especially considering that our garage is in the basement and our house is four stories tall, including the partial daylight basement. The app accurately shows even what side of the garage the car is in. My phone usually has trouble picking up a GPS signal in the garage, but the car does not.

A few weeks ago, while I was driving, the GPS did misidentify our positon by about a half-mile. At one point, it thought we were on the runway at Dobbins AFB when instead we were on a state highway. The GPS lady was very confused; I half expected her to warn me of a C-130 on final. I rebooted the screen several times and it finally figured out where we really were about 10 minutes later, on I-75. I haven't seen this problem since.
Exactly what I was referring to. You were probably on South Cobb, or possibly the interstate and the GPS was returning a point that was actually closer to the runway than whichever road that you were on. It was demonstrating the "pinning" of itself to a road (or runway in this case) and guessed wrong. It will tend to stay lost until it finds a reason to recalculate it's position, i.e. you make a turn on a road, or the road runs out.
 

Ed Woodrick

Well-Known Member
Joined
May 25, 2018
Messages
277
Location
Atlanta, GA
Tesla Owner
Model 3
Country
Country
#13
Guess I've got some good street data for my neighborhood, my car is perfect in the garage every time I've ever checked. Even where my partner parks at work I can look it up and know exactly which parking spot he got at work that day..
A parking spot is actually a lot easier to be more accurate on. GPSs, because of the accuracy problem, will tend to average out a fixed location over time. The average of positions over time is pretty accurate.

In the case of DGPS, I believe that you can leave your reference receiver in the same location for a few hours and it will then provide relatively accurate differentials to other nearby receivers.
 

Ed Woodrick

Well-Known Member
Joined
May 25, 2018
Messages
277
Location
Atlanta, GA
Tesla Owner
Model 3
Country
Country
#14
We have several GPS-based NTP Stratum 1 clocks at work (so we know the time to within sub-microseconds). The antennaes cost ~$75/US or so. They will give a location to within 0.1 feet, aka 1.2" with latitude, longitude, and altitude .I realize the electronics in motorcars isn't and doesn't need to be to that level.

Would be nice if I can "move" the GPS "center" location in the car 20 feet or so so it gives the correct address.
Yep, stationary devices, as in my previous post, can get pretty accurate.
 
Joined
Jun 25, 2017
Messages
125
Location
San Luis Obispo
Tesla Owner
Model 3
Country
Country
#15
In good conditions (open sky) single GPS receivers can have an absolute accuracy of 1-10 feet. Where GPS is very accurate is in relative positioning (relative to another receiver). That's why differential GPS is better. Static, differential GPS can have accuracies measured in millimeters.The other sources of error in this system are the mapping and the addressing data, which I suspect have more error than the GPS in favorable conditions. In hostile GPS environments (adjacent to buildings, under tree canopy, etc.) the satellite signal multipath seriously reduces GPS accuracy.
 

evannole

Active Member
Joined
Jun 17, 2018
Messages
55
Location
Marietta, GA
Tesla Owner
Model 3
Country
Country
#16
Exactly what I was referring to. You were probably on South Cobb, or possibly the interstate and the GPS was returning a point that was actually closer to the runway than whichever road that you were on. It was demonstrating the "pinning" of itself to a road (or runway in this case) and guessed wrong. It will tend to stay lost until it finds a reason to recalculate it's position, i.e. you make a turn on a road, or the road runs out.
Good detective work! I was on South Cobb, just west of Fairground Street. I am rarely going that way anymore, now that the toll lanes are open!