I recently surveyed an area using Reach RS2 as a base for a RTK drone system. The base was set up as Average Single accumulating for 2 minutes to build a relative-accurate map. After a while I got the opportunity to get coordinates of points on the ground that could be easily identified on the orthomosaic generated from the drone views.
When comparing my map with the absolute-accurate points I measure an offset of 16m. This seems a bit much given that emlid website advertises 2.5m absolute accuracy with this method. What do you think ?
Plus, I have done some testing on the ground without flying my drone to see if the 15m offset could come from this, and I only got a 15 cm offset beetween using average single method vs average fix.
If you could enlighten me on this topic this would be great !
That’s also my opinion. However I cannot explain what else this offset could result from as RTK was connected with the drone’s reveicer and broadcasting corrections.
The area was an electric transformation station. Maybe the Emlid RS2 error in average single can skyrocket near electrical installations ?
The offset is uniform on the entire map (I placed 4 checkpoints to control results).
Indeed, electrical networks can also cause disturbances which lead to degradation of accuracy.
In such cases, try to lengthen the observation time and make recordings.
My advice is to always log at the base, especially when using a drone for mapping. The data can be useful in resolving error issues and also prevent a second trip to the site.
I would only use Single if I knew I would postprocessing the location later.
Average Fix, 5-10 mins should be fine, if you don’t plan to PostProcess after. But really depends on your Baseline.
Why not use your local base instead of NTRIP ?
So base for the UAV and GCP’s were not the same? Any practical reasoning behind that?
Sounds right, and still you would probably not get anything better than 2.5m vertical. Atmospherics change all the time, and you don’t get any correction for that.
Agree, 5-10 mins still isn’t a lot for a typical static reference , but vastly(!) better than 5-10 minutes of Single.
Practical reason is that I can only use one RS2 receiver. So I use it as a local base for my mapping mission for the drone and then (or before) I use it as a rover to collect GCPs.
Fort both uses I get a fix solution receiving corrections from a NTRIP provider within (10-30km max).
Following you advice I will increase the averaging time up to 10-20min even for average fix mode.
I always collect 1min average fix solution in the emlid base.
Leaving the unit for 10-20min in average fixed mode will not give you much better solution especially if there is open sky view. 5min would be a good occupation time for average fixed.
After “Base launched with new coordinates” message is displayed then I turn off the incoming NTRIP corrections and the receiver returns to Single mode.
My problem is that if the single solution is not close to the fixed solution then the base doesn’t function properly and the rover doesn’t fix (not even float) and stays single.
Yesterday I did a test and collected a point in fix mode and then stakeout the same point in single mode. The horizontal was ok within 2m and the elevation was off by 5m.
Other test gave me better results within 1.5m.
I came to this conclusion because if I let the base station in fixed mode everything works fine on the rover side.
If it is possible on emlid to increase the threshold of the difference between the fixed coordinates and the single solution on the base mode I think would solve the problem.