Someone posted on our OSM Colorado Meetup group stating that their Tesla Model 3 built in navigation is routing them in a less than ideal manner. I have checked and OSM seems to be correct and reasonably complete. The issue seems to concern this section of 14th Street / CR 14 1/2 in Fort Lupton Colorado: Way: 14th Street (1426397619) | OpenStreetMap
The issue is that when routing from the south, such as Brighton Colorado, the Tesla algorithm has them exit US 85 at 14th Street/CR 14 1/2 in order to get to their neighborhood, which is around this way: Way: Willow Xing Parkway (1333705257) | OpenStreetMap The issue is that part of 14th Street is unpaved, and they prefer to avoid this.
They believe exiting US 85 at 1st Street ( Way: 1st Street (665664327) | OpenStreetMap ), is a better navigation choice. I have assured them that OSM mapping of 14th Street and 1st Street are reasonably correct, and that the issue is likely Tesla. Either the Tesla routing algorithm or the frequency of their data updates. Does anyone insight as to how the Tesla algorithm works, or how frequently they update their copy of OSM (or any other thing that may be a factor)? It is interesting to note that none of the example routers on the OSM homepage choose 1st Street in this routing scenario.
After doing some research on Reddit, YouTube, and a few Tesla forums, it seems to me that Tesla relies on Google Maps for map display. When it comes to the routing engine, I’ve read two conflicting opinions. One post mentions Azure Maps or Bing Maps and thus TomTom (which also uses OSM), while another post claims that Valhalla is used for routing.
However, I was unable to generate navigation that goes through 1st Street in either routing engine.
Maybe it’s related with this post and it has to be fixed on Tesla’s side. Maybe you can contact navfeedback@tesla.com to get the issue fixed.
I went through and updated the area a bit more, so it should be as up to date as we can get it right now. That connection between the County Road 14 1/2 and County Road 29 corner and the Willow Crossing development looks like it’s unofficial and shouldn’t exist honestly (but it does, so it’s mapped).
I think the issue is on the router’s side, and probably has to do with a “limit the number of turns” type of feature. Is there a way for them to avoid unpaved roads (like a box to check or something)?
Also they could just ignore the Tesla GPS until it adjusts the route properly, but that’s just me being squirrely.
You will find dozens of threads on the topic even just in this forum.
The most likely source of Tesla nav data is Tomtom though there is no official statement on the matter.
In any case it is not OSM or Google.
This is easy to show in areas where the navigation is out of date (by years) and calculated routes contain roads that no longer exist or use to have significantly different geometry (the Google part of this is immediately visible as the displayed map -is- from Google).
Thanks everyone for your replies and your research. I have invited the person that posted on the OSM Colorado Meetup to this discussion, but will also summarize there for them.
@SD_Mapman Thanks for the updates in the area! I am in Colorado, but seldom make it over to Fort Lupton.
As a mapper and routing nerd who also owns a Tesla Model 3 with FSD, I keep a close eye on how it picks routes. A few observations may be useful for anyone in the OSM community who’s curious about Tesla’s map stack.
The in-car UX is clearly Google-based. The base map, turn arrows, and general presentation track closely with Google Maps, with some Tesla styling layered on.
In day-to-day driving it behaves in ways that don’t line up with current OSM data. If Tesla is using OSM at all, it’s either a very old snapshot or limited to very specific layers.
The car navigates indoors without GPS, including multi-level garages that neither OSM nor Google have mapped. That appears to be pure dead-reckoning and sensor fusion rather than map data.
There is a recurring claim that Tesla uses OSM parking-lot roads. That might be true in some markets, but Apple Maps (TomTom-heavy, plus OSM) also has excellent parking-lot coverage, so it’s hard to know whether any OSM influence is direct or via a data partner.
One thing I can say with confidence: improving OSM will not affect Tesla routing in the near term. I have a daily route where the car insists on going through a gated, private road. It’s correctly marked as private/closed in OSM, and both Google and Apple avoid it. Tesla still tries every time, even after multiple “report map issue” submissions from the car.
Tesla’s routing remains something of a black box. As a tech-minded driver it’s fascinating to watch, even when it makes baffling choices.
I would note, and I’ve pointed this out elsewhere before, there is naturally no rule that Tesla needs to use the same nav data source for all markets. With other words we can’t rule out that there might be markets in which they do.
But given that they don’t use OSM in DACH it would seem nonsensical to assume that they use it any where else where OSM is not as good.