GPS Blending with RTK

Hi all,
I have lurked on the forum for a while now - a great bunch of people and knowledge on here.

I have an S800 DJI frame converted to using a Pixhawk 2.1 FC and Here GNSS mounted on a post about 15cm above the top deck - so far I have been absolutely delighted with the performance. I normally operate it with a 32bit Alexmos gimbal onboard and take oblique images for photogrammetry.

I now have a need for much better positional accuracy both relative and absolute and nadir only images. I have just added a custom nadir Sony camera mount with triggering by a Seagull Map2 from the Pixhawk and hot-shoe feedback to an onboard Emlid M+ with antenna mounted directly above the camera optical axis, minimising lever-arm displacement (although I am working on calcs from the Pixhawk IMU data to calculate these). All systems are working well on the ground (weather has been shocking lately, so not had a chance to test fly). I have for the Emlid, RTK working well from a base station feed to the copter via RTK injection through Mission Planner.

I currently have GPS blending set and is working well on the ground. My question is, if the Emlid looses RTK fix or if one antenna fails, will there be a significant jump in positioning to cause confusion of the EKF and cause maintenance of attitude or positioning issues in auto/loiter/pos. hold/RTL modes? I know recent releases have addressed the jumping issue to a degree (I’m running 3.6.7 Chibios), but I have searched the forum and can’t find a situation where someone has tested a RTK fix solution GNSS blended with a non-RTK.

BTW when developing and building the craft, I found it hard to find relevant information on motor numbering, expected / realistic PID values for the frame and other areas - so happy to share any such information with others if they need it - just PM me.

Best regards


1 Like

Blending will probably not work correctly because the GPS are from different vendors, and therefore the way they measure and report accuracy numbers is different.

Blending code relies on the accuracy numbers being related, but that is only the case if both GPSs come from the same vendor.

Thanks for reply. I think I’ll play it safe and just use the Here gnss for navigation and leave the Emlid purely for photo geotagging.

Can you point me to which parameters the blending uses to determine accuracy - just interested in seeing the difference between the two.