Development of an Autonomous Navigation System using a Two-dimensional Laser Scanner in an Orchard Application

The objective of this study was to develop an automatic guidance system capable of navigating an autonomous vehicle travelling between tree rows in a real-time application. The study focused solely on straight line recognition of the tree rows using a laser scanner as a navigation sensor. A 52 kW agricultural tractor was used as the platform on which the laser scanner was mounted. A Hough transform was used as the algorithm to recognise the tree row. An auto-regression method eliminated the white Gaussian noise in the laser scanner data. A calibration method was used to select the offset position of the laser scanner and to correct the heading and lateral error evaluation. An appropriate speed for tractor was also determined. By obtaining an accuracy of 0·11 m lateral error and 1·5 ° heading error, it was possible to navigate the robot tractor autonomously between the orchard row crops.