| Target tracking intelligent car technology fusion computer visualization,AI,autogenous control and other hot techniques in many fields,has a broad application prospect,in the advanced follow luggage,follow golf bag,intelligent follow shopping cart,storage intelligent handling car,and other scenes are widely used,these application scenes generally on real-time,flexibility,practicality requirements will be higher,in addition in the visual tracking Accuracy and stability are also constant problems that cannot be ignored.In this paper,we study the control methods of tracking two aspects of the operation of intelligent vehicles from the viewpoint of actual implementation,The first is to obtain the target’s position in real time through a visual tracking arrangement and convert it to a realistic coordinate system;then the corresponding control methods are used to control the cart to move to the corresponding position according to the coordinate position obtained by the visual method in order to follow the target movement.First,for the available goal following algorithm either the precision is not high,in complex background application anti-interference ability and detection tracking fast enough;or practicality is not enough,the calculation is complex and computationally large,relying on the upper computer map transfer to complete the control operation in real time is not good.This paper proposes the use of the internal color features of the follow-up object as detection features to allow automatic acquisition of the tracking subject;and fuse the external shape features of the target to make improvements to the calculation program to increase the tracking accuracy of the check and following calculation program when the target is obscured or moving too fast in complex backgrounds,while the computational burden of the calculation program is not greatly increased;in addition,this paper still uses the Kalman movement state estimation to forecast the coordinate of the object in the video This paper also uses Kalman movement state estimation to forecast the coordinate of the object in the next frame to increase the accuracy and speed of detection.Secondly,for the problems of stability and following the target object in real time during the control of the intelligent vehicle,PID control using the fuzzy control method is proposed to eliminate the oscillation and inaccurate motion control of the trolley motor and servo during operation and improve the stability of the servo rotation;The result of fuzzy control is used as the input of DC electric motor PWM controle to improve the response of smart car movement controle,reduce the stable state mistake,and achieve smooth and fast smart car control.Finally,on the basis of the above theoretical and algorithmic research,the tracking algorithm program is designed and written using VS2019 programming software with Open CV4 algorithm library,and the modeling of the cart control is carried out using MATLAB,Simulation tests are conducted on the vision part and drive part of the intelligent car respectively to validate the feasibleness and improvement effect of the modified tracking arrangement and test the PID control of the fuzzy controller on the cart motion The system control performance is improved by PID control with fuzzy controller.The core control board is designed based on STM32h743VIT6 microcontroller,and the executable control program of the microcontroller is compiled and debugged by IDE and burned to the microcontroller;the OV7670 image sensor,L298 N motor driver board and other hardware devices are used to build the crawler vision intelligent vehicle,and the implementation of the software and hardware system of the intelligent vehicle is completed,the practicability and effectiveness of the visual smart vehicle controlling systems devised in this article are validated. |
PDF Full Text Request