Use heading from NMEA when available to set picture direction

[Eric_S]

When correlating images with GPX track, one can ask to use GPX track to set image direction (Set Image direction towards the next GPX waypoint). Indeed, such direction can be provided by the GNSS chip itself, for instance when used with an Inertial measurement unit (IMU). I'm using a ublox F9R which combine a RTK GNSS receiver with an IMU. Heading is outed in several NMEA sentences (RMC, VTG and THS). By using such data, one can set the right direction when driving backward but, more importantly, avoid random directions when being stationary (at traffic lights). Extreme case, my camera was 2 m in front of the GNSS antenna. I indicated this in JOSM while doing the correlation. Pictures were placed on a 2 m radius circle around the GPS point when waiting at traffic lights.

$GNRMC,074949.50,A,4511.1929356,N,00544.6631314,E,0.001,**292.66**,270724,2.32,E,R,V*4C
$GNVTG,**292.66**,T,290.35,M,0.001,N,0.001,K,D*3C
$GNGGA,074949.50,4511.1929356,N,00544.6631314,E,4,12,0.67,217.148,M,47.399,M,1.0,0000*62
$GNGLL,4511.1929356,N,00544.6631314,E,074949.50,A,D*73
$GNTHS,**292.66**,A*10

[StephaneP]

I think that trying to use values from the VTG and RMC sentences isn't a good idea, because it's not the heading value, but the course over ground. THS seems not very common.
Does your setup output HDT values? Don't you think It looks like a better candidate?

[Eric_S]

Yes, VTG and RMC are supposed to be course while THS is really heading. For a car, they should be equal only when IMU and antenna are above the middle of the rear axle. Indeed, with ublox F9R, they are all equal in any case. But, for a generic approach, I agree that one should focus on heading.

HDT is supposed to be obsolete: ​https://gpsd.gitlab.io/gpsd/NMEA.html

HDT should be replaced by THS.