We have received many questions about the real accuracy of the simpleRTK2B board with the ZED-F9P module and we will write a series of posts talking about it.
The accuracy of a GNSS receiver depends on many factors, e.g. antenna used, vehicle dynamics, view of the sky, …
So every time you see an accuracy figure, you should ask under which conditions it has been measured.
In this test we wanted to focus on accuracy and repeatability so we prepared a special test bench:
- Base and rover simpleRTK2B boards are prototypes
- Passive GNSS multiband antenna for base and rover.
The final antennas (both OEM and IP65) will be active, with better signal reception.
- Drill turning at a fixed trigger position, with a speed ~38rpm
- Distance (radius) from GNSS antenna to drill center of 60.96cm
- GNSS antenna speed of ~2.3m/s
- simpleRTK2B working in base-rover configuration (RTK)
- simpleRTK2B configured to report data at 5Hz
- 41.7 minutes of data (12510 points), including start and stop of the drill.
If someone is curious, the complete log of the test can be downloaded here.
- All the results shown have no post processing/filtering.
Test results, showing relative position wer base (blue) and real position (red):
For each of the 12510 points, the error wrt the real position has been calculated and this is its histogram PDF and CDF, fitted with a Weibull distribution:
The accuracy is:
- <7mm (~1σ, 68% of the points have an accuracy better than 7mm)
- <1.7cm (~2σ, 95.45% of the points have an accuracy better than 1.7cm)
- <3.8cm (~3σ, 99.73% of the points have an accuracy better than 3.8cm)