Dynamic Target Capture Of The Mobile Robot Based On Extended Kalman Filter

With the increasing market of mobile robots,a large number of mobile robots appear in the environment of logistics transportation,shopping mall service,airport handling,etc.,and the requirements of human for the positioning,control and tracking performance of mobile robots are increasing.In the positioning and motion of mobile robot,the traditional Kalman filter is a linear system in the positioning process,which needs a lot of observation data to accurately predict and can not be used in the complex nonlinear environment.Extended Kalman filter is an extension of Kalman filter,which can solve the problem of non-linear motion,and only need a small amount of observation data to accurately locate the target,and then through the combination of robot control and path planning method can quickly capture the target body.In this thesis,two kinds of location prediction methods,mean approach method and extended Kalman filter location method,two path planning methods,the current shortest path method and the shortest path method after motion are studied.The main contents of this thesis are divided into two aspects:(1)The extended Kalman filter(EKF)and mean value approximation(MA)are used to locate the mobile robot.The extended Kalman filter mainly studies the state evaluation through the moving target motion model to predict the coordinates of the next time point of the moving target,and then the observation evaluation based on the observation model.The data association between the prediction evaluation and the current observation is established through the extended Kalman filter,and the predicted target coordinates can be quickly located by updating with the Kalman gain.The mean value approach can simulate and fit the linear velocity and angular velocity of the moving target through a large number of observation data,and realize the positioning prediction of the moving target.(2)In the process of mobile robot acquisition,the path planning algorithm is studied.Two methods,the current shortest path and the shortest path after motion,are proposed.The current shortest path selects the direction of the nearest moving object according to the distance of the mobile robot to multiple moving objects.The shortest path after moving calculates the total distance of the mobile robot for multiple moving objects according to the last step of the mobile robot moving towards each moving object,and the mobile robot will move towards the direction with the minimum total distance.The experimental results show that the combination of the extended Kalman filterand the current shortest path method has the most advantages for the acquisition of multiple dynamic targets.