Noggrannhet av RTK-GPS-positionering av enskilda frön vid precisionssådd

Sammanfattning: With higher demands on farmers to reduce the usage of fertilizers and pes ti cides because of their negative environmental impact, as well as the rising prices of resources required for agriculture, precision agriculture has become a big focus in today’s agriculture research. Individual plant care systems are a big part of precision agriculture, and accurate information about the po si tions of individual plants is required for these systems. In collaboration with RISE Research Institutes of Sweden and Väderstad AB, a seed mapping system for the planter Väderstad Tempo was developed. The system was designed to determine the individual seed positions by receiving geographical coordinates from a real time kinematic global positioning system (RTK-GPS) in con junc tion with signals from the seed sensor already installed on the planter. A sec on dary task in the project was to isolate the error sources which would af fect the final error of the positioning. Variation in the GPS system’s reported coordinates as well as variation in the seed fall time from the seed sensor to the ground were investigated. Seed fall times were measured in laboratory conditions where the type of crop, air pressure in the seed housing, and angular velocity of the seed disc were varied in order to build a mathematical model of the seed fall time. The de via tion metric root mean squared error (RMSE) of maize, sunflower, sugar beet, and cereal were 2.97 ms, 4.58 ms, 1.72 ms and 3.23 ms respectively. Measurement of the variation of the RTK-GPS system was conducted in conjunction with the final tests of the system in a field. On average, the latitude and longitude had a RMSE of 4.17e-8 degrees and 6.38e-8 degrees respectively, which corresponded to a RMSE of 4.6 mm northing and 4.0 mm easting at the specific location of the measurements. The seed mapping system was tested in a field under operation velocities in the range 5-15 km/h with the crops maize and cereal. After the seed mapping process, the seeds were found and positioned in a combination of RTK-GPS and manual measuring in order to determine their actual position. In the planter's driving direction, it was found that a seed could be found within a distance of 62.34 mm of a mapped seed position with a 95% probability. The errors investigated for the seed fall times and RTK-GPS measurements turned out to only account for a small part of the total error of the seed mapping system. The remaining error stems from other parts of the process and needs to be investigated further.