Research On Visual Trajectory Tracking System Of Wheeled Inverted Pendulum Robot

In recent years,the demand for robotic visual trajectory tracking such as unmanned vehicle driving,image recognition,and tracking intelligent systems has become more and more high,which has attracted the attention of many people.The vision-based robot trajectory tracking system has the characteristics of multivariable,strong coupling,nonlinearity,autonomy and so on.Therefore,it is of theoretical and practical significance to study the trajectory tracking problem based on vision.A visual trajectory tracking system for inverted pendulum robots is designed,which combines visual control with trajectory tracking control to guide the robot to the desired trajectory.Firstly,the multi-instance target tracking method(MIL)is introduced for the case where the classifier causes tracking drift due to inaccurate classification.Multiple training samples are placed in a training package,and the training classifier is obtained by labeling the training package to achieve the purpose of classifying and predicting the test package.Secondly,in order to achieve the purpose of tracking robot navigation guidance,a method of tracking of the line of sight is proposed.A dual closed loop PID controller is designed by acquiring heading and distance information to extract the desired trajectory of the tracking robot.The trajectory tracking controller is designed by the trajectory linearization method to obtain an open-loop inverse system and acquire the required input according to the required output.Adjust the error in a closed loop system.Then,the final input of the system is obtained according to the expected input of the open loop and the closed loop error input.Finally,the above two parts are designed to realize the visual trajectory tracking system,and the results are analyzed.The experimental results show that the multi-instance target tracking method can track the target in a short time even if the target has occlusion,which indicates that the method can effectively solve the tracking drift.By combining tracking of the line of sight with the linearization of the trajectory,the decoupling of the system and the adjustment of the error are realized.The controller can control the robot to track the target trajectory,which indicates that the designed controller is feasible.