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

Tesla Autopilot maps

This site may earn commission on affiliate links.
No doubt the Tesla parking lot was mapped to that extent for the FSD demo.

That is possible but not necessary. There are more parking lots mapped that way. I looked at some tiles to the north of Tesla HQ, and this is how Stanford Shopping Center looks like (sorry for the poor quality this time):

ShoppingCenter.jpg


So regardless how the creation of this maps is done, this kind of details are not put there just for the purpose of one demo.

I don't know how the maps are made, and I should not speculate abut it, but let's not reject right away the concept, that the maps are (at least to some degree) created from the data upload by the cars. Again, I don't know that, but if it's true, that it would be somehow normal, that the places visited by Tesla cars more often are more precisely mapped.
 
Please enlighten us @DamianXVI how you produce those black lines on that map....

Right now it looks like this:
  • I'm finding Geohash of the place using an online decoder.
  • I'm downloading the tile files using a web browser.
  • Then processing this tile using a piece of code in C# that was created as a result of my experiments with the file format, and that right now is a horrible hackish mess with lots of hardcoded paths, and stuff...
  • This code is creating an image, draws bounding boxes of geohashes found in it and on top of it draws spline paths (using Bezier curves) from the points read from the file.
  • Next I'm grabbing a screenshot of Google maps and composing it with the image of paths in Paint.NET. Sometimes stitching a couple screenshots together. First images were combined using geohash boundaries as reference (to be sure that the roads actually match by their coordinates, not by me stretching them), last images I just combined so that the roads match with the map.
I believe this is not the full answer for your question... Right now I don't have much time, but I do intend to work on this more in a couple of weeks, and maybe (but just maybe) create some tool or viewer.

In the meantime, have a piece of San Francisco Bay Area.
 

Attachments

  • big_map2.png
    big_map2.png
    754.6 KB · Views: 317
That is possible but not necessary. There are more parking lots mapped that way. I looked at some tiles to the north of Tesla HQ, and this is how Stanford Shopping Center looks like (sorry for the poor quality this time):

View attachment 260215

So regardless how the creation of this maps is done, this kind of details are not put there just for the purpose of one demo.

I don't know how the maps are made, and I should not speculate abut it, but let's not reject right away the concept, that the maps are (at least to some degree) created from the data upload by the cars. Again, I don't know that, but if it's true, that it would be somehow normal, that the places visited by Tesla cars more often are more precisely mapped.

Agree - was thinking more of the feature lists... Any chance of a dump of the feature lists for the Tesla parking lot?
 
Hm, I found what looks a little bit like a tile format in the ape code too, but I am not sure it's the same thing because I don't see any curves or speed limit in there.
It also only defines two furniture types: "stop line" and a "sign' (traffic sign?).

Stop lines seem pretty significant. That seems to suggest there's at least a path they're trying to pursue (ugh that's an awful pun, I'm so sorry) to the car knowing where to stop if it starts understanding stop signs and traffic lights….
 
This doesn't sound very safe. Even assuming that the maps are 100% perfect, what about stuff that isn't on them? New roads, alterations, temporary road surfaces, private roads, roadworks...

Seems like the only safe option is to measure the curve optically. Otherwise you get into Apple Maps levels of badness.
 
  • Like
Reactions: Kant.Ing
Any chance of a dump of the feature lists for the Tesla parking lot?

No, I'm not able to give any definitive answer right now. A single path record can have at least 27 fields, I figured out the meaning of 6. There is still a lot of work to do.

I tried to look however at the data of record defining this loop closest to the entrance, an at first glance it does not look to be significantly different when compared with other records from other tiles, but I can't say for sure.
 
Thanks for this discussion guys, I have asked Tesla about the car slowing going into corners and was told that it doesn't do it! I drive on rural roads in the south of Australia and was puzzled by what was providing the info for decision making in TACC. On a section of road I drive frequently the car was slowing down for one sharp curve, but not for a second further along. After the last upgrade (or independently, may have been coincidence) it no longer slows "at all" for the first curve, but now slows heavily prior to the second curve and is doing that consistently for the past month. I look forward to consistency but to some degree the element of surprise with constantly changing behaviours makes life interesting.
 
Korea gets its own tiles for some reason.

South Korea is very sensitive about maps. There's a NYT article that explains some of it.

Travelers who want to go from Gimpo International Airport to the Gangnam neighborhood of Seoul cannot rely on Google Maps. Google Maps can provide directions only for public transport, not for driving, to any place in Korea. Anyone crazy enough to try the journey on bicycle or on foot, directions for which Google Maps provides elsewhere, will be similarly stymied.

South Korean security restrictions that were put in place after the Korean War limit Google’s maps, the company says. The export of map data is barred, ostensibly to prevent it from falling into the hands of South Korea’s foe to the north, across the world’s most heavily fortified border. Google and other foreign Internet companies say the rule also prevents them from providing online mapping services, like navigation, that travelers have come to rely on in much of the rest of the world.
Most things to do with street level maps either leave Korea out entirely, or are forced to handle it very differently. I know one system will download maps as soon as it's sure you're in South Korea, and immediately delete it if you leave the borders.
 
China is the same. Google maps only have major roads in many areas, and they are often wildly inaccurate and don't match up at all with satellite photos. The only reliable source of maps for China I have found is Baidu.
 
Next are at least two lists of IDs referencing other paths, probably pointing to connecting paths, forming a big graph of routes.

Next are more lists (at least five) with some features, that I didn't figured out the meaning of, so there are more things in this map.

Latitude and Longitude are stored as 32bit unsigned integers, mapped to the whole range of a given value (fixed point notation).

Do you see any indications that it contains objects along the paths that are the geocoded whitelist of objects that the radar can see (overhead higway signs, overpasses, bridges etc)?
 
Do you see any indications that it contains objects along the paths that are the geocoded whitelist of objects that the radar can see (overhead higway signs, overpasses, bridges etc)?

Yes.

Thanks to the help from verygreen I know, that one of the fields in the spline path entry is related to "radar". This field looks to be a list of sections along the spline length. That is, it is defined based on the start point and the end point along the spline (where 0.0 is the beginning of the spline, and 1.0 is the end of the spline). That means those are not just points, but sections of the road.

Each section has two more fields. One is an integer in range 0...100, that looks to be some kind of a confidence level, but I'm not sure. Other is most likely an elevation readout reported by the radar (because it looks similar to the value that can be found in the .inputs file in the snapshots).

I tried to find those places on the streetview. Sometimes it is not clear what is special in a given point, in other cases it is very clear. Below couple examples of some more clear situations:

Link
Link
Link
Link
Link

What is interesting is that it looks like quite often those places are mapped because of the trees. Either trees growing very close to the side of the road, or with branches over the road. Like in this place:

Link