| With the introduction of the concept of smart agriculture,agricultural production is gradually developing in the direction of industrialization and intelligence.Because the internal environment of the cotton field is complex and the distance between the fields is narrow,it is not conducive to manually entering the cotton field to check the situation.Therefore,this research studies the control technology of the intelligent vehicle that can be used to help the perception of the cotton field environment.This technology is based on machine vision,combined with fuzzy control method and PWM speed control system to control the autonomous driving and steering of smart cars.This study mainly has the following three aspects:（1）Identification and enhancement of local path images in cotton fields.Because the path between cotton fields is an unstructured road,it is necessary to perform image preprocessing,segmentation and morphological enhancement algorithms when extracting cotton field path information to enhance the path information in the collected images and improve the efficiency of intelligent vehicle route extraction..In image preprocessing,the dynamic Gaussian homomorphic filter based on HSV color space is used to compensate the illumination of the collected local cotton field path images,so as to weaken the influence of illumination factors on cotton field path recognition.Then,the path image converted back to RGB color space is gray-scaled by the super-green eigenvalue method,so that the cotton plant and the path can be better separated.Median filtering is used to suppress and eliminate the noise in the image,and then the maximum inter-class variance method is used to separate the path of the image from the background.Finally,the path information in the image is enhanced by morphological filtering.（2）Extract the navigation lines in the cotton field path.Firstly,the enhanced cotton field path image is extracted by the contour search method and the least square method to make the path edge coherent and clear.To meet the requirements of real-time and accuracy in the driving process of smart cars,a navigation line extraction method based on the centroid method and the least squares method is proposed.Curve fitting,the fitted curve is the automatically planned navigation line during the driving process of the smart car.（3）Control the smart car to track the navigation line.Using the fuzzy control algorithm and PWM speed control system,the intelligent vehicle is controlled to follow the path and steer along the extracted route.Select two points on the left and right sides of the navigation line of the smart car as the input of the fuzzy controller,formulate the control principle of the fuzzy controller,optimize the PWM duty cycle of the motor,reduce the lateral difference between the smart car and the navigation line,and make the The smart car can drive automatically along the navigation line accurately.The camera of the smart car is calibrated,and the navigation line tracking control experiment of the smart car is carried out.In the experiment,the error of the smart car deviating from the navigation line is small,which shows that based on machine vision,combined with fuzzy control algorithm and PWM speed regulation system,the smart car is controlled to drive autonomously in cotton fields.with steering feasibility. |
