Foliar Fertilizer Robot for Raised Bed Greenhouse Using NDVI Image Processing System

Abstract

This research aims to construct and test a foliar fertilizer robot that determines the volume of fertilizer spraying based on the Vegetation Index (VI). The robot, which was designed to move in a cartesian mechanism, was tested with Green Oak lettuce plants on a raised bed, size of 1 x 6 x 0.25 m (W x L x H) in the greenhouse, size of 10 x 20 x 5 m (W x L x H). The average VI values of fertilized and watered Green Oak lettuces in low and high light intensity were 0.577 and 0.564, respectively. As the result, the lettuces without fertilizer and water for 3 days have decreased to 0.544 and 0.546, respectively. The camera set up in the distance of 45 cm was the most appropriate distance because the VI values were relatively sTable. The height of the nozzle at 40 cm shown the appropriate volume of fertilizer solution at 4.8–9.2 ml/plant and the optimum spraying time based on plant health level was determined at 0.1, 0.7, 1.4, and 2.0 seconds because spray volume was consistent with the academic principle. The linear regression model from spraying at 20 cm high nozzle shown the highest accuracy at 97.62% and the lowest error with MSE and MAE at 0.0326 and 0.1503, respectively. Fertilizer wastage compared to spraying with the robot and a person with a backpack sprayer and greenhouse fertigation system was 0.27, 0.41, and 1.67