# The Tunnels GPS Problem, a Solution Approach

**URL:** <https://forum.mapillary.com/t/the-tunnels-gps-problem-a-solution-approach/9283>\
**Category:** Contributing and equipment\
**Created:** [February 20, 2025, 9:19pm UTC](https://forum.mapillary.com/t/the-tunnels-gps-problem-a-solution-approach/9283 "2025-02-20T21:19:03Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![osmplus\_org](https://sea1.discourse-cdn.com/flex015/user_avatar/forum.mapillary.com/osmplus_org/32/72465_2.png) [@osmplus\_org](https://forum.mapillary.com/u/osmplus_org)\
**Post date:** [February 20, 2025, 9:19pm UTC](https://forum.mapillary.com/t/the-tunnels-gps-problem-a-solution-approach/9283/1 "2025-02-20T21:19:03Z")

</div>

Tunnel issue: In a fully enclosed tunnel there is no reception of navigation satellite signals. It is therefore difficult to create a correct GPX trace without additional vehicle telemetry data.  
Tip: Video recordings have a separate GPX file. This can be opened with the OpenStreetMap JOSM editor and converted into an editable JOSM path using the contact menu. You are now not allowed to remove or add any GPX points. Distribute existing GPX points with the JOSM key combination Shift + B, you may have to resolve Luups. Then distribute the points evenly across the tunnel. You can export the way as GPX using the JOSM editor. The result will vary, but I find this satisfying.

Example: [Mapillary](https://www.mapillary.com/app/?lat=47.258061534400014&lng=11.431497445100035&z=19.9&panos=true&pKey=628691183443413&x=0.27147162610323944&y=0.5782954728240529&zoom=0&focus=photo)

---

<div class="post-metadata">

**Author:** ![boris](https://sea1.discourse-cdn.com/flex015/user_avatar/forum.mapillary.com/boris/32/71767_2.png) [@boris](https://forum.mapillary.com/u/boris)\
**Post date:** [February 21, 2025, 12:05pm UTC](https://forum.mapillary.com/t/the-tunnels-gps-problem-a-solution-approach/9283/2 "2025-02-21T12:05:19Z")

</div>

Yes absolutely. You can also edit the .gpx in a number of other tools like [www.gpx.studio](http://www.gpx.studio)

---

<div class="post-metadata">

**Author:** ![bob3bob3](https://sea1.discourse-cdn.com/flex015/user_avatar/forum.mapillary.com/bob3bob3/32/68393_2.png) [@bob3bob3](https://forum.mapillary.com/u/bob3bob3)\
**Post date:** [February 22, 2025, 3:17am UTC](https://forum.mapillary.com/t/the-tunnels-gps-problem-a-solution-approach/9283/3 "2025-02-22T03:17:13Z")

</div>

Might even be worth snapping it to the OSM way as well. ie create a GPX track for (say) the kerb side lane (overpass?) and make the vehicle run at a fixed speed in that lane. Mapillary tools allows an absolute start date/time, but exiftool could just write a date/time agnostic EXIF. Have to make sure the way geometry is correct though, eg from the original construction data.

---

<div class="post-metadata">

**Author:** ![osmplus\_org](https://sea1.discourse-cdn.com/flex015/user_avatar/forum.mapillary.com/osmplus_org/32/72465_2.png) [@osmplus\_org](https://forum.mapillary.com/u/osmplus_org)\
**Post date:** [February 22, 2025, 7:11am UTC](https://forum.mapillary.com/t/the-tunnels-gps-problem-a-solution-approach/9283/4 "2025-02-22T07:11:42Z")

</div>

Running a vehicle at a fixed speed is unrealistic in the reality of public transport. Another approach could be helpful here. In areas where there is no GPS reception, vehicle telemetry data could be used. There are countless apps that record such data using an ODB2 Bluetooth adapter. There is usually already an OpenStreetMap route for the route.  
In such a case, the following four files are required:

1. The video recording from the camera for the **image** information
2. Useful **time** information is already contained in the video or photo.
3. The ODB2 Car data for the **speed**
4. A GPX file for the **route** created from OpenStreetMap data

A solution that would work for the average Mapillary contributor would be to rely on all clocks being synchronized, and to offer OBD 2 vehicle data as an additional upload if available.

---

<div class="post-metadata">

**Author:** ![Eric\_S](https://avatars.discourse-cdn.com/v4/letter/e/aca169/32.png) [@Eric\_S](https://forum.mapillary.com/u/Eric_S)\
**Post date:** [February 22, 2025, 9:06am UTC](https://forum.mapillary.com/t/the-tunnels-gps-problem-a-solution-approach/9283/5 "2025-02-22T09:06:51Z")

</div>

Use a GPS coupled to an IMU (inertial measurement unit) like in the ublox F9R and you get that:

> **[Mapillary](https://www.mapillary.com/app/?pKey=1007764011040636)**
>
> Street-level imagery, powered by collaboration and computer vision.

I was also considering getting vehicle speed from ODB2 and transmitting it to F9R for better results (one can see a glitch at tunnel outlet). I was able to get the speed of my car but I didn’t merged the code with the one of my [F9R logger](https://github.com/Eric-FR/F9x_RAWX_Logger).

Right now, I’m more considering smoothing the NMEA position using NMEA speed/dynamic, the latter one not showing glitch at tunnel outlet.
