2007 Minneapolis, Minnesota, June 17-20, 2007 · 2007 · 17 citations · 0 references
EngineeringAutonomous SystemsRobotics EngineeringAgricultural RoboticsAutonomous VehiclesRoboticsAgricultural Machine VisionAutonomous Field NavigationAutonomous NavigationVehicle PerceptionYaw DynamicsAutomated Row FollowingAutonomous Farm VehiclesAutonomous Vehicle ControlDead ReckoningIn-row NavigationInertial Measurement Unit ControlTurningTurning Tracks
This article discusses the development of turning maneuver for an autonomous vehicle from a row into the adjacent row at the headlands of citrus groves. The objective was to navigate the grove alleyway in the center, turn into and navigate the headland, turn into the adjacent row and start navigating the adjacent row. A turning maneuver is executed in a series of five steps which are 1. Determining the approach of the end of current row. 2. Determining the end of the row. 3. Turning into the headland. 4. Turning into the next row. 5. Getting to a position to navigate the next row. A John Deere eGator was used as the autonomous vehicle. The sensors used for executing the maneuver are Video Camera, Laser Radar (ladar), Inertial Measurement Unit (IMU) and wheel Encoder. The video camera and ladar aid in detecting and navigating the headland. The yaw angle measured by the IMU provides the turning angle and the wheel encoder is used for speed control and dead reckoning to confirm end of row. Row navigation performance of the autonomous vehicle was tested in test tracks. The maximum error was 3.5cm for an average path width of 2.5m. Row navigation performance and headland turning maneuver were then tested in a citrus grove. The vehicle navigated the alleyways in the center of the path between the trees with a maximum error of 9 cm for an average path width of 3.5m. The vehicle turned into the headland at the end of a row, navigated the headland and turned in to the adjacent row without any contact with the alleyway trees and headland trees. Headland turning ability combined with row navigation ability enables an autonomous vehicle to completely navigate a citrus grove.