Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Abstract. Machine vision was one of the main sensor technologies for agricultural robots to detect the working environment and objects. Determining the distance between the actuator and the target was the basis for accurate control. Among common methods, monocular distance measurement had the advantages of low cost and fast calculation. Aiming to develop the monocular distance measurement model for the picking robot of pomelo fruit, an imaging system based on one low-cost industrial camera was designed and set up and the relationship between the change of imaging distance and the difference of target fruit pixels in the image was analyzed in this work. Under the same lighting conditions, the image data of checkerboard calibration boards with various sizes and specifications were collected from the forward horizontal direction at each 5cm spacing within 1.5 meters. The pixel value of the checkerboards in the images under different shooting distance conditions were calculated by using Photoshop software. Then the mathematical relationships between distances and pixel values of multiple checker groups with different area sizes were assessed in MATLAB. Basing on the above relationships, the monocular distance measurement model was established according to the regression analysis results. The absolute error of distance detection was less than 0.025m, and the average error was less than 3%, which indicated that combined with mechanical design and control parameter optimization, the distance measurement based on single camera would be qualified for guiding the picking actuator.