DEVELOPMENT OF AN AUTONOMOUS NAVIGATION AGRICULTURAL ROBOTIC PLATFORM BASED ON MACHINE VISION

Abstract

In Egypt, management of crops still away from what is being used today by utilizing the advances of mechanical design capabilities, sensing and electronics technology. These technologies have been introduced in many places and recorded high accuracy in different field operations. So, an autonomous agricultural robotic platform (ARP) based on machine vision has been developed and constructed. The ARP consisted of two main parts namely; 1) Power transmission and auto-guide system; and 2) Robotic platform. The experiments were carried out at department of agricultural engineering, faculty of agriculture, Kafrelsheikh University during 2014-2015. In this study, the experiments were conducted in laboratory to optimize the accuracy of ARP control using machine vision in term of the autonomous navigation and performance of the robot’s guidance system. For evaluation the image processing technique, four different camera resolutions (1080×1920, 1944×2592, 2736×3648, and 3240×4320 pixels) and three camera' heights (500, 700 and 1000 mm) have been used to measure the execution time for image processing steps. Flight time of spray droplets has been calculated under three levels of spray height, (70, 100 and 130 mm), three levels of spray pressure (1, 3 and 5 bar) and three levels of nozzle size, (1.5, 3 and 5 mm). Also, the effect of changing duty cycle percentage (DC, %) has been studied to control the speed of the ARP with Pulse Width Modulation (PWM) signals. Different nozzle tip sizes and spray pressures have been used to trace the flow rate variation. Based on the total time consumed in the execution time and droplets flight time, speed of the ARP has been noted according to the resolution of camera and its height levels. Results showed that the robotic platform' guidance system with machine vision was able to adequately distinguish the path, resist image noise and give less lateral offset error than the human operators.The average lateral error of autonomous was 2.75, 19.33, 21.22, 34.18 and16.69 mm, while the average lateral error of human operator was 32.70, 4.85, 7.85, 38.35 and 14.75 mm for straight path, curved path, sine wave path, offset discontinuity, and angle discontinuity respectively. The best execution time of image processing was obtained with the minimum values of the camera resolution at 500 mm camera height. While, increasing the size of nozzle at same height and spray pressure decreased the flight time. The favorable robotic platform' speeds were obtained at lower values of camera resolutions and wider distances between nozzle and camera.