Navigation of the spraying robot in jujube orchard

Abstract

This study presents a solution for achieving high-precision cruising along a predefined operational path to enable fully automated spraying in a densely planted jujube orchard. A fully autonomous navigation system for a jujube orchard spraying robot, based on a combination of LiDAR SLAM and IMU inertial navigation, was designed. The system integrates IMU and LiDAR positioning information using an extended Kalman filter algorithm. The navigation system uses LiDAR to detect the orchard environment and perform SLAM mapping, while the AMCL algorithm determines the robot's position in the map using sensor localisation data. The LiDAR-detected data between rows in the orchard is clustered and fitted to extract operational points between the rows. The robot's spraying path is designed using the A* and DWA algorithms, enabling specific path cruising for the spraying robot in the high concentration of orchards with jujube plants. Experiments conducted in the high concentration of orchards with jujube plants show that when the robot travels along a 15-meter path, the mean deviation in the X-axis is 3.41 cm, and the average yaw angle is 1.25°. When moving within the 1.5-meter fixed-point parking area, the mean X-direction deviation is 1.98 cm, with an average yaw angle of 0.77°. When the robot turns with a radius of 2 m, the average deviation in the X-axis is 1.17 cm, the average distance deviation parallel to the trajectory is 4.79 cm, and the average yaw angle is 3.31°. Additionally, when the robot follows an S-shaped path, the mean deviation in the X-axis is 1.8 cm, and the average yaw angle is 1.6°. The system meets the cruising requirements for actual plant protection spraying operations. It offers high navigation accuracy, providing an effective reference for autonomous navigation in densely planted jujube orchard spraying operations.