Silvilaser 2019 - Poster Presentations »
Estimating charcoal piles volume using an unmanned aerial vehicle
A common challenge for charcoal production companies is to implement inventory control in the context of Manufacturing Resources Planning (MPR). Large charcoal production is conducted in rectangular ovens with a capacity of around 320 cubic meters of wood. However, the carbonization process is not statistically controlled, and the production is accounted in the MPR considering an average value per cycle. During the charcoal shipment, the trucks are weighted and the volume is gravimetrically computed. The consequence of the difference between the precision of the shipment discount (weighted) and the accreditation of the production (average) in the MPR is a permanent divergence between real and MPR charcoal stock. The objective of this study is to compare the volumetric estimation of charcoal piles based on digital stereoscopy to the estimation based on GNSS/RTK field points. Two charcoal piles were surveyed using GNSS/RTK and their volumes were topographically estimated. In the same day, an aerial image collection was performed with DJI Phantom 3 Standard. The flight planning was performed in the software Pix4D Capture, using a double grid design, and flight height of 40 meters. The resulted ground sample distance was 1.86 cm/pixel. The images were processed in two ways: low accuracy for photos alignment and for dense cloud creation; and high accuracy for photos alignment and for dense cloud creation. HIgher the accuracy slower the processing time. The piles' polygon were visually created based on the orthomosaic and the volume determined based on the digital elevation model. The volume determined for the first pile was 461 m³ and 328 m³, for the higher and lower accuracy respectively. The second pile had 185 m³ and 135m³ the higher and lower accuracy respectively. The volume determined by GNSS/RTK survey were 399 m³ and 175 m³ for piles one and two respectively. The low accuracy processing resulted in lower volumetric model quality. As a consequence, the analyst had greater difficulties to delineate the piles and less precision in the solid creation. The values found by the high accuracy processing approximate more to the GNSS/RTK values. For piles monitoring using UAVs, we recommend processing considering high accuracy parametrization.