Research On AGV Path Tracking And Obstacle Avoidance Strategy Based On Vision Guidance

AGV(Automated Guided Vehicle),as an intelligent guidance device with 24 hours continuous operation,can replace workers to realize the unmanned process operation.The rise of digital image processing technology and machine vision has promoted the development of AGV vision guidance technology.In addition,vision guidance has the advantages of flexibility and low cost.Aiming at the problem of vision-guided AGV running according to the preset path,this paper conducts path tracking control research.The controller is designed to improve the sampling time lag,optimize the fuzzy reasoning and realize the online self-tuning of parameters,combining the simple and easy realization of PID control algorithm,the reasoning ability of fuzzy control algorithm and the efficient learning ability of neural network algorithm.The controller designed in this paper has stable performance and good rectifying performance.At the same time,design obstacle avoidance strategies,formulate obstacle avoidance safety areas,establish obstacle avoidance models and ranging models to realize AGV avoiding obstacles without collision.Finally,the model is established with Prescan,and the timeliness and stability of the AGV fuzzy neural network PID controller are verified based on four groups of path tracking simulation experiments.The specific research content and results are as follows:(1)Firstly,the collected images are preprocessed,and the RGB data format images are converted into grayscale images.The improved adaptive median filter is used for noise reduction and enhancement processing,and the guide line and obstacles are separated from the background based on the adaptive Ostu threshold segmentation algorithm.Canny operator is used for edge detection based on image features.The Hough change is used to identify the boundary line of the path and fit the center line.The obstacles are detected and marked with a rectangular box.(2)Design obstacle avoidance strategy.Formulate obstacle avoidance safety zones,and establish three obstacle avoidance models based on the position of the obstacle relative to the guide line.Establish the longitudinal ranging model and the lateral ranging model of the obstacle relative to the AGV,based on the pinhole imaging model.The camera calibration and ranging experiments are carried out to verify the effectiveness of the ranging model designed in this paper.Meanwhile,the AGV obstacle avoidance simulation is carried out based on the obstacle avoidance model.(3)Adopt guided-driven AGV body structure to establish the corresponding kinematics model.Based on PID algorithm,this paper designs PID + feedforward controller to improve the problem of sampling time lag on traditional PID controller,combines fuzzy control algorithm to design fuzzy PID controller to solve the problem that PID can not adjust parameters online,and integrates neural network algorithm to design fuzzy neural network PID controller to optimize Fuzzy logic reasoning.The corresponding control system model is established by adding noise interference,and the simulation and comparative analysis show that the fuzzy neural network PID controller designed in this paper is not easy to cause oscillation,meanwhile that has fast response speed and good stability.(4)The 3D model of the AGV car body and the 3D scene model are established,and the AGV path tracking simulation experiment is carried out with Prescan.The tracking control verification was carried out for the linear path,the circular arc path,the linear transition arc path and the curved path respectively.The results show that the AGV fuzzy neural network PID control system designed in this paper can track the path quickly under the interference of noise signal,and its robustness,timeliness and control stability are ideal.