Development of a Combined Orchard Harvesting Robot Navigation System

Our research concerned the development of an autonomous robotic navigation system for orchard harvesting with a dual master-slave mode, tractor transport robot being master followed by picking as slave. This addresses problem that in single agricultural combined equipment cannot stop repeatedly between rows apple trees and drive continuously when turning. According to distances obtained from global positioning (GNSS), ground points were used switch mode robot. A cloth simulation filter (CSF) random sample consensus (RANSAC) algorithm was obtain inter-row waypoints. The GNSS point manually selected turn waypoint kinematic model compute waypoints Finally, we pure pursuit track these sequentially achieve head command navigation. experimental results show data packet loss rate less than 1.2% communicated row within 50 m which meets communication requirements. can repeated stops using follow navigation, demands joint harvesting. maximum, minimum, mean standard deviation position 5.3 cm, 0.8 2.4 0.9 respectively. deviations slave larger those robot, 39.7 1.1 4.1 5.6 following error 4.4 0 1.3 1 cm Concerning turn, method allowed continuous turning, but lateral robots more 0.3 m, heading 10° 90°.