Hello fellow mappers,
I am very excited because of three things. I like OSM and Mapillary, I drive a Tesla model 3 and I have learned that there are ways I can use the in build cameras of my Tesla to record and upload to Mapillary.
There is a Mapillary blog post from 2020 and a few post in this forum.
Before I dive head first into this adventure I wanted to ask around again if there are any follow ups to the blog post (anywhere on the net) or if there are other users that can share some of their experience with this undertaking?
I am curious to read what body of knowledge is out there!
From my side I have started this project. My first steps are to figure out mobile phone GPS recording and āTesla Dashcam operationā so that they donāt get in the way while driving.
I have opted to try https://gpslogger.app/ on my phone. Any other suggestions are welcome.
My first impressions are:
Lots of options, maybe too many
saving of files needs to be changed so that it is not stored by day but per recording session
accuracy reported by the app seems bad (around 4-5 m most of the time)
On the Tesla Dashcam side I have formatted by USB stick and are now trying to figure out if I need to press the recording button every 10 minutes or just to turn it on or off. The manual seems to indicate that I need to save every <10 minutes.
Once saved, Dashcam saves the last 10 minutes of footage.
Great use of those cameras.
Iāve seen a few images from what appeared to be vehicle positioning cameras, but donāt think there was a comprehensive write up.
I use teslausb on a raspberry pi, to keep a rotating ~100G archive of tesla footage. Very cool project, it emulates a usb stick but prevents the video from being deleted.
I have some scripting that checks when the raspberry pi becomes visible on wifi, and starts uploading the video-files to mapillary.
I customized mapillary_tools to:
Not upload a sample for a location if thereās still data less than 3 months old
Not upload a sample if itās less than 5m away from the previous sample.
I should really do a proper write-up and release these modifications properly. In the meantime; Iād be happy to help you get setup. Ask me if you get stuck anywhere.
BTW; The scripts from the blog-post are incompatible with current versions of mapillary_tools. This is my script: teslamapillary.py Ā· GitHub
But make sure to alter the video_sample_interval. Itās currently set to 12fps which is fine since I hacked the tooling to wait until thereās 5m between two samples, but for the default tooling itās better to stick to 3-5fps or so.
Very cool projects. BTW @seb - you donāt have to use mapillary_tools to add gpx files to videos, this is now natively supported in the Mapillary Desktop uploader (overflow menu next to the video file will let you select the associated gpx).
wow, this is really awesome and much more skilled that I am able to for now.
I read about the teslapi but donĀ“t have the time at the moment. will put it onto the whish list.
@boris, thanks for pointing this out. I did try to use the desktop uploader but get an āUpload failedā error. I had a few tries but all without success. made separate forum post.
Hey @TumbleCow , would you be able to help me with the python script a bit.
do you point the script to a folder with all cameras in (front, sides, back) or manually move them around?
What format is the time zone in? UCS+11, AEST or AEDT (for Australian Eastern Daylight Time) all result in an error
OK Australia/Sydney worked!
do I have to point the script to individual video files or to a folder?
do I need to point it to a GPX file or a folder with GPX files in?
the script seems to have issues with my mapillary email. Mine is in the form name+mapillary@gmail.com
how can I hand usename and password to the script more permanently?
$ python3 teslamapillary.py --help
usage: teslamapillary.py [-h] mapillary_user timezone gpxpath videofiles [videofiles ...]
Process Tesla dashcam videos with accompanying GPX file
positional arguments:
mapillary_user Mapillary username
timezone Timezone the Tesla ride was in. Default: Europe/Zurich
gpxpath The gpx file
videofiles The videofiles to process