ERM LLC recently conducted a survey of a housing development construction site using three different DJI small unmanned aircraft systems (sUAS), the Phantom 4 Pro, the Phantom 4, and the Phantom 3 Pro. ERM used the Pix4D capture app in 2D (parallel rows) and 3D (cross hatch) flight planner.
Ten Aeropoints were placed throughout the 30 acre site as shown below. The five large marks were used as ground control points (GCP) and the 5 small marks were used as checkpoints (CP). The average variance of the 10 points, as reported by Propeller, was 0.45, 0.36, and 1.26 centimeters (cm) in X, Y, and Z respectively.
Some of the critical parameters from thePix4D quality reports on each mission are shown in the table below. Clearly, the 20MP camera on the P4 Pro provides superior survey accuracy, as compared to the 12MP cameras on the P3 Pro and P4. The P4 3D survey grid did not cover the southernmost GCP well causing high RMS errors in that survey. This GCP was switched to a CP, the CP to the NNE of it was switched to a GCP and the survey was reoptimized in Pix4D. The results are shown below (2) and resulted in much a much better quality report.
|sUAS and Pix4D Grid||Checkpoint Avg. RMS Error (cm)||GCP Avg. RMS Error (cm)||GSD (cm)|
|P4 Pro 2D||0.136||0.668||2.03||0.244||1.50|
|P4 Pro 3D||0.746||1.17||3.29||0.549||1.28|
|P3 Pro 2D||4.77||3.40||9.57||3.17||2.52|
|P4 2D (2)||8.16||4.94||0.46||4.40||2.15|