Checkpoint Geolocation Issue

Your assistance in resolving this is appreciated. Processed a map with GCP’s and CP’s. The GCPs and CPs were completed in the same workflow, same rover (Emlid Reach RX). GCP RMSE is as expected, CPs are completely scattered, even though they were done by the same individual and tagged by the same person. Can you help sort this out?

Even after tagging several times - quite accurately, the resulting ortho does not show the CP’s where they should be. Here’s some screen shots -
The intersection of the “X” is the proper location. The processed map placed them in a completely different location.

Where they were tagged - the corner of the storm drain grate.


1 Like
  • Looks to be about a foot?
  • What is the relationship of the marker to the visual? Always off in the same direction/amount?
  • Can you export the full points report from Reachview? That would give us more data to see the health of the receiver during the shots.
  • Where were you getting your corrections from?
  • How far was the baseline?

Michael, thanks for jumping in on this. The RMSE for GCPs is 0.51". The CP RMSE is 4.93". They were captured within the same workflow, immediately after the GCP collection. My interpretation of the issue is DD’s AI is ignoring where I tag the CPs and using the ground targets to factor what it thinks the geolocation is. The ground targets were not used in the typical manner - a permanent geographic landmark was used adjacent to the ground target. Examples - the corner of a storm drain grate, the corner of a grave marker, etc. The ground target was only placed so the aerial image could identify the general area, then tag the actual CP. This method was used in the event the RX had issues (which has occurred in the past) and we could return and identify the location once the ground targets were removed (also took ground images of the precise pole placement).

Corrections were from an NTRIP base just a few miles away. Attached is the detailed export.

I’ll reply again with the Map report RMSE pages.

![Walnut Hills Preliminary Report-Copy_1|386x500](upload://tXVMTNBD18nHZDCciGLJgLHRm3J.jpeg

DD does a lot quite well, but this policy of one attachment for “new users” really sucks.

Overall the points report looks pretty good. I am left wondering why some numbers appear to be manually entered as the default seems to be rounding to the 1/1000th but there are also values rounded to the 1/100th and 1/10th?

On the last report it definitely looks like something was awry with the tagging. Almost like you tagged the spot you wanted and the processing used the marking? I would definitely get with support to have them take a look. Can you place a pin at each location and compare the DroneDeploy reported value against the Emlid value?