Design And Implementation Of Orchard Weeding Robot Control System Based On FPGA

In traditional orchards,large mechanized weeding can reduce labor but its control system is less intelligent and can easily damage fruit trees in the process of operation.Designing a targeted intelligent control system is costly and difficult to develop.Therefore,the mechanical automation and intelligence of orchard weeding has become the development direction of digital orchards in line with the new era.Intelligent control system is an important measure of the degree of mechanical intelligence,the existing automatic orchard weeding machinery control system has high power consumption,weak expansion and development capability,and low real-time performance.This paper addresses this problem and designs a weeding robot with field programmable gate array(FPGA)as the main control chip,taking into account the topographical features of mountainous orchards in China.The weeding robot is equipped with an image processing system,a weeding system,an auxiliary navigation system and a travel power system for weeding between monopolies in orchards.In this paper,the FPGA is used as the control core,and the FPGA control chip has the advantages of low power consumption,high expansion capability,high real-time and parallel processing capability,etc.The image processing algorithm and sensor data fusion algorithm are transplanted to increase the intelligence of the weeding machine and reduce the development cost and difficulty.In this paper,we use vision system to collect image information and generate path navigation line by image processing algorithm,and fuse the data from gyroscope and laser rangefinder sensors to control the weeding robot for autonomous inter-monopoly navigation.The details of this paper are as follows:(1)Based on the topographical conditions of mountain orchards,the composition of the weeding robot is determined: light crawler chassis,electric drive system,FPGA master controller,vision system,gyroscope and laser rangefinder multi-sensor navigation system,and the location of the sensors is reasonably arranged according to the orchard environment.(2)Complete the construction of hardware structure and function configuration.The corresponding hardware structure is designed according to the image system function,where the image system function mainly includes the data reading function of the camera,data storage function,and image data processing function.The configuration of each hardware structure includes: the configuration of OV5640 camera,the configuration of memory,and the configuration of VGA display function.(3)Complete the hardware transplantation of image processing algorithm.According to the processing flow of image algorithm,the image algorithm is divided into bottom,middle and top layer algorithms,and the algorithm is hardware ported and the algorithm functions are verified by test experiments,including bottom layer image pre-processing algorithm: conversion of original color image and Y-Cb-Cr space,binarization and noise reduction of grayscale image;middle layer image algorithm: feature point extraction of feature area;top layer algorithm: feature point The navigation line fitting.(4)Complete the autonomous navigation function of the weeding robot between orchards.The image based navigation line generation,fusion of gyroscope and laser range finder sensor data,adjustment of the real-time attitude of the weeding robot,and field experiments in the orchard monopoly to complete the weeding robot orchard monopoly weeding work and achieve the function of obstacle avoidance for large obstacles in the orchard monopoly.The orchard field experiments show that the weeding robot control system designed in this paper has good image processing ability between the monopolies,and the maximum lateral error of straight line walking between the monopolies is 0.03 m,which basically meets the obstacle avoidance function for large obstacles between the monopolies.