Development of high-speed camera hardware and software package to evaluate real-time electric seed meter accuracy of a variable rate planter

•Average singulation of 98.45% when planting at speed varying from 2.4kph to 16.1kph.•The seed metering system exhibited no triple seeds and only 0.03% instances of doubles.•Planting at seed rates above 1250 seeds/min increased singulation errors.•Seed singulation errors were twice for speed transitions of 4.8kph/s as compared to 2.4kph/s. Electric drive seed metering systems have become common for planting row-crop seed to accommodate increased machine size and planting speeds and to allow individual row unit-control that enable site-specific planting for spatially sensitive areas and contour farming. Seed singulation (a measurement of singulated seeds, misses, and multiples) is critical requirement when adopting high speed planting. However, current planting controllers fail to indicate whether singulation errors occurred due to operator-based behaviors such as speed changes, headland operation, point rows and contour farming at varying speed transitions (accelerations/decelerations). Therefore this study was conducted to understand a seed metering system’s ability to singulate seed under typical scenarios with specific objectives to (1) quantify electric seed metering accuracy using high-speed imaging and (2) identify machine operating states that impact seeding accuracy. A Horsch Maestro 24.30 planter was sent commands to plant at constant speeds of 7.2, 9.7, 12.0kph while accelerating/decelerating at 2.4 and 4.8kph/s from/to a stop and between speeds. The planter was sent commands to plant around contours at varying radii (20, 40, 80, 150m) at varying speeds (i.e., 0, 2.4, 4.8, 6.4, 7.2, 9.7, 12.1, 12.9, 14.5, 16.1kph). Simulations were conducted at two rates (44,550 and 89,110seedsha−1). A high-speed imaging system was developed using LabVIEW to record real-time seed meter singulation at 300frames/s by combining planting machine states with seed tube sensor data and vision based seed measurements to quantify single count seeds, misses, and multiples. When planting from 2.4kph to 16.1kph, results showed an average singulation of 98.45% where errors nearly doubled with fast accelerations and decelerations and abrupt changes such as a shift during headland turns. Overall, planting above 1250seeds per minute resulted in an increased number of singulation errors. The vision based measurements were within 0.8±0.2% of the commercial seed tube sensors. The seed per minute value which provided optimal seed singulation can be used as a control parameter by tec