Abstract:Automatic navigation operation has become an inevitable trend as the development of agricultural machinery and equipment. To realize the sprayer working in field with the minimal manual intervention and avoid the pesticide injury to human in manual operation, on the basis of electrification transformation of the spray machine, Lovol high clearance boom sprayer ZP9500, the automatic navigation operation system was developed based on RTK-GNSS (real-time kinematic - global navigation satellite system). Electro-hydraulic steering system with a proportional flow control valve was equipped on ZP9500, and an angle sensor was coaxially connected to steering knuckle arms, realizing the closed-loop control system for steering. For the same advantage to throttle system, a push-rod electric machine with travel sensor was installed, and the push-rod was connected to accelerator via wire rope. Installed in parallel, 2 electromagnetic relays were connected to the electronic switch of vehicle clutch and spray pump respectively, and 3 electric two-way valves were connected to 3 spray bar switches. So, the steering, accelerator, vehicle travelling, spraying and sprinkling width of sprayer could be operated in electric control. GNSS receiver with double satellite antennas and inertial sensor component, were applied in the system as the crucial position and orientation measurement equipment. Based on the hardware platform, the path planning strategy, automatic navigation controller and automatic operation controller were developed. The automatic navigation controller could guide the vehicle along the planned path generated by path planning strategy, and complete straight line tracking and headland turning. The curve tracking controller was designed based on pure pursuit model, and the linear path tracking controller was designed based on states feedback of lateral error, yaw angle and speed. And the operation controller could realize the vehicle start or stop, spraying, amplitude adjustment and speed switch control automatically according to different working conditions,such as RTK signal out of order, vehicle headland turning, and spraying operation completed. To verify the reliability and accuracy of the automatic operation system, at the speed of 1.3 m/s, the test was carried out in the field of cement pavement, dry farmland and paddy field. Due to the difference of field size and setting width, 4 rows were planned in cement pavement, 5 rows were planned in dry farmland, and 4 rows were planned in paddy field. Test result showed that: 1) The navigation operation system could automatically guide the spraying machine to complete the spraying operation in field, the switch between straight line tracking and curve tracking was smooth, and all the functions performed reliably. 2) The tracking accuracy and stability, and lateral position deviation were consistent in the same farmland, but in different conditions, it showed obvious differences. On cement pavement, when swinging on rough road, the vehicle rolling angle was from ?1.6? to 1.5?, the maximum tracking error of the lateral direction was 3.9 cm, the average error of each row was from ?1.0 to 0.4 cm, and the standard deviation of deviation of each row was 0.8-1.4 cm. In dry farmland, the vehicle rolling angle was from ?1.4? to 3.3?, the maximum tracking error was 9.8 cm, the average error of each row was from 0.9 to 1.9 cm, and the standard deviation of deviation of each row was between 2.1 and 4.6 cm. In paddy field, the vehicle rolling angle was from ?2.4? to 5.2?, the maximum tracking error was 17.5 cm, the average error of each row was between 1.4 and 2.9 cm, and the standard deviation of deviation of each row was between 3.0 and 5.2 cm. 3) By studying the tracking error and the rolling angle of each row, the maximum tracking error had significant negative correlation with the maximum rolling angle, and their negative correlation coefficient was 0.914. Meanwhile, the standard deviation of tracking error and the maximum rolling angle had the same characteristic relationship, and their negative correlation coefficient was 0.947.