Research And Development Of Agricultural Robot Test Platform Visual Navigation System

In recent years, the state has increased the support to agricultural mechanization and automation, which greatly promoted the development of China’s agricultural modernization. Agricultural robot played a pivotal role in the development of modern agriculture. In this paper the autonomous mobile robot vehicle is a kind of wheeled farmland vehicle, visual navigation system of the autonomous mobile test platform can simulate wheeled robot visual navigation, facilitate the research and development of the farmland robot visual navigation system. This paper focuses on agricultural robot test platform visual navigation platform, studied the robot visual navigation image process methods, designed robot visual navigation path tracking strategy and navigation controllers, mainly include the following work:(1) Overall design of the agricultural robot test platform visual navigation system. Mount a camera on the agricultural robot test platform as the image acquisition module. The professional image processing system DSP used to process the digital image and calculate navigation parameters. Eventually navigation parameters were transmitted to the STM32 MCU control system for motor control. The agricultural robot test platform equipped with ultrasonic sensors to detect obstacles, enabling the mobile vehicle to avoid obstacles.(2) Agricultural robot test platform visual navigation test platform camera calibration. Build the camera imaging mathematical model using the pinhole model to get the camera inside and outside parameters, and verified the calibration results by experiment.(3) Study the popular visual navigation image processing method for farmland road image process. Through the study of popular visual navigation image processing method select the reasonable image processing method, and ultimately determined a suitable image processing algorithms.(4) Put forward a new tracking preview point path navigation method and obtain path navigation parameters through modeling. After studying the farmland Monocular Vision Navigation general method is proposed based on a new path along the edge of farmland traveling robot vision navigation strategy, the establishment of a farm autonomous mobile geometry relationship between the vehicle and the navigation route, through the construction the actual position of the navigation module for solving preview point, according to preview the final point, navigation path edges and farmland autonomous mobile relative position of the vehicle to strike a path to navigate between the three parameters.(5) Establishment of kinematic model of the agricultural robot test platform, according to the navigation strategy design navigation system controller on the basis of agricultural robot test platform wheel motor operating characteristics. Firstly, through the analysis of the kinematics model of agricultural robot test platform to understand the kinematics characteristics, and the vehicle wheel motor operating characteristics were studied to understand the dynamics of agricultural robot test platform. Based on the agricultural robot test platform navigation strategy, select appropriate input and output parameters to design a navigation controller, design PID- Fuzzy integrated Control by the utilization of classic PID controller and fuzzy controller.(6) Agricultural robot test platform visual navigation system software design and experiments. Software design mainly includes DSP image process programming and STM32 MCU System control programming. DSP programming includes a series of image processing to determine preview point and obtain the navigation parameters. DSP transmitted the calculated navigation parameters to the microcontroller control system, microcontroller used the navigation parameters as controller input parameters, controller outputs control agricultural robot test platform wheel motor to achieve control aims, namely track navigation path. Finally, simulation experiments verified the feasibility of agricultural robot test platform visual navigation system by simulating farmland environment.