Bresearch On Vision-based Navigation For Agricultural Vehicle

With the increasing aging trend of population and corresponding farm labor shortage in the world, the development of precision agriculture and agricultural mechanization is the necessary way to solve the shortages of food production and labor. The continuous progress of science and technology has prompted the rapid development of agricultural equipment. Intelligent and automatic equipments used in agricultural machine have become a hotspot in agricultural engineering. Autonomous navigation is one of key technologies. Developing agricultural vehicle autonomous navigation system which is simple operation, low cost, and high precision, is significant to agricultural production. On the one hand, autonomous navigation can improve the operation quality, reduce the driver’s labor intensity, and improve production efficiency; on the other hand can, it can improve the work condition, avoid pesticide poisoning to human body, reduce the investment of agriculture productivity, decrease energy consumption, save cost, and increase agricultural economic benefits. This paper investigated the basic structure, control strategy, and design scheme of hardware and software of the navigation system in detail.Firstly, it is necessary to reconstruct the original steering system of the tractor for autonomous navigation. According to the structural characteristics of the original tractor steering, and the reformation plans of domestic and international tractor auto navigation, the control scheme of electronic hydraulic steering was determined. A simulation steering model based on PD control algorithm with dead zone was established by SimHydraulics software. The experiments which were conducted on the real tractor show that, the electronic hydraulic steering has good characteristics in response and stability and preciously control the steering angle of wheels according to the commands from PC.Secondly, according to the characteristics of field crops and background environment, the plan of visual information processing was determined. The visual program of auto navigation for agricultural vehicle was compiled by Matlab software. This program could extract crop-line (i.e., expectation path) via the images of the road in front of the tractor. And then it could obtain the angle and position deviation between tractor and crop line. On the basis of deviations, PC sent desired angle which was controlled the tractor on the right path to electronic hydraulic steering.Then, steering controller was designed and made. It can accept PC desired instructions and real-time angle sensor information, and send out the control command to the electric hydraulic steering. This controller includes the smallest system, proportional and directional valve control circuit, communication circuit and angle sensor acquisition circuit etc. Control program was written based on the demands of the system and controller hardware.Finally, in order to verify the system of vision-based auto navigation for agricultural vehicle, using Maoyuan-250 as the platform to do some experiments. The test results showed that, the vehicle speed was 2.08km/h, the front wheel steered rapidly and stably, the maximum error was 0.18m, and the average error was 0.048m. The system could guarantee the agricultural vehicle to realize the auto navigation in the field.