Study On Self-Balancing Robot Control System Based On Vision Guidance

With the characteristics of small volume,simple structure,flexible movement and small radius of motion,the two-wheeled inverted self-balancing robot has unparalleled advantage as a mobile carrier in the research of service robots and small ball motion robots.In recent years,visual tracking technology has developed rapidly,which has become an important branch in the research of mobile robots.It will greatly improve the autonomous capability of self-balancing robots through combining the control technology of the two-wheeled self-balancing vehicle with the visual navigation technology.This paper aims to explore the three-dimensional motion modeling,control algorithm design,visual guidance and hardware implementation of two-wheel self-balancing robot,which lays the foundation for further research on the robot with service function and small ball motion.First,the dynamic analysis of the self-balancing robot is discussed based on the Lagrangian method,the three-dimensional dynamic equation of the self-balancing robot is established,and these works provide a comprehensive foundation dynamic model for the following control algorithm research.Then,in order to keep the self-balancing upright and stable,the backstepping controller design under the known model condition and the unknown model adaptive condition are discussed respectively based on the Lyapunov stability analysis.Based on the self-balanced control,the steering control based on PD controller and the speed controller based on PI controller are discussed.By simulation of the controller,the correctness and validity of the corresponding controller algorithm has been proved.Next,the method of target recognition based on color feature is discussed.A color histogram of the target region is established by using the single channel of the HSV color space.The gray image of the probability distribution of the color of the subject is obtained by using the backward projection algorithm,which realizes the segmentation of target and backgrond.Camshift target tracking algorithm achieves the recognition of continuous images object by setting the target position of the current image as the starting point of the next image for Meanshift search.According to pin-hole camera imaging model,the position of the tracking target is estimated,and the trajectory of the car is adjusted by point stabilization method to realize the tracking of the target.Finally,according to the demand of self-balancing robot,the control system hardware platform based on STM32F103 is constructed.The motion drive module realizes the driving of the motor.The electronic gyroscope module realizes the detection of the body posture.The wireless communication module realizes the data transmission between the equipments.The camera module realizes the target image tracking.Through debugging,the basic motion control function is realized.