Research On Control Method Of Vehicle Based On Vision Guidance

With the development of automation technology and the launch of the "Industry 4.0" strategy,mobile robots based on computer vision technology are gradually playing a role in many fields.According to different application scenarios,different mobile robots are derived.For example,self-driving vehicles,automatic guided vehicles(AGV),automatic cleaning vehicles in shopping malls and office buildings,automatic disinfection vehicles in medical places,etc.,the application field of robots has gradually expanded from the industrial field to the service industry [1].This paper takes wheeled mobile robots as the research object,and conducts the following researches on two vision-based guidance methods:(1)By analyzing the movement principle of the Mecanum wheel vehicle,the forward and inverse kinematics model of the vehicle is given.(2)In order to solve the problem of incomplete information extraction and low accuracy in the navigation band,a series of processing methods based on HSV color model,nonlinear filtering,Otsu threshold processing method,Canny edge detection and line fitting are proposed to get the navigation line equation.It is verified by subsequent experiments that this method can segment the navigation belt more accurately while ensuring a certain real-time performance.From this,the position and attitude deviation of the vehicle relative to the navigation belt is calculated,and the linear speed and angular velocity of the vehicle are controlled to achieve the effect of correcting deviation,so as to control the vehicle to drive along the navigation belt autonomously.(3)A way to guide the vehicle through target tracking is designed.The scale estimation method is introduced into the framework of the kernel-related filtering algorithm to make up for the lack of a fixed size of the tracking frame of the traditional algorithm.This paper first uses the characteristics of the camera model and the depth camera to make a preliminary scale estimation,and then uses the method of establishing a scale pool to make more precise scale selection.Subsequently,based on the depth information,through the relative distance between the vehicle and the tracking target and the horizontal position of the center point of the tracking frame,the linear velocity and angular velocity of the vehicle are calculated and updated in real time to ensure that the vehicle can accurately and smoothly follow the target.(4)In the actual scene,the two guiding methods of the vehicle moving along the navigation belt and following the tracked target are verified.Then experimental results tell us the two guidance methods can achieve accurate guidance control for the vehicle.