Baseline detection and matching to vision-based navigation of agricultural robot

An automatic guidance model based on machine vision for detection and localization of crops rows is presented. The machine vision system consists of a color video camera and a computer. The camera is mounted on the head directly above the robot as a navigation sensor. When the agricultural mobile robot goes forward, the camera captures images continuously and transferred to the computer. First, pattern recognition and image processing were used to obtain quasi navigation baseline. Second, the real navigation line was extracted from quasi navigation baseline via Hough Transform. Then the mobile robot can be guided by the navigation line matching dynamically by itself. Test results indicate that the model has simple and robust algorithm, low-level requirements for software and hardware and was capable of navigating an agricultural autonomous robots traveling between crops rows.