Study On Fixed Target Tracking And Obstacle Avoidance For Mobile Robot

Mobile robot target tracking has become a popular research in the robotics field,it has a large value both in the military and civilian applications.The ability to avoid obstacles automatically when the mobile robot encounters unexpected obstacles during target tracking is an important basis for realizing automatic target tracking for a mobile robot.This research bases on the study of the previous research in-depth,a fixed target tracking and obstacle avoidance method for mobile robot which gears to 2-D flat surface and 3-D rough ground was proposed.In the meanwhile,an experimental platform for the fixed target tracking and obstacle avoidance of a mobile robot was designed,and the method was demonstrated to be accurate through the target tracking and obstacle avoidance experiments on the 2-D flat ground and 3-D rough terrain.The main work of this paper is organized as follows:Firstly,according to the task requirements,the overall mobile robot target tracking and obstacle avoidance scheme was proposed.In the modular design,the system is divided into visual tracking module,obstacle avoidance module and motion control module.Based on the scheme design philosophy,an experimental platform consisting of a visual tracking unit,an obstacle avoidance unit and a control unit was built.In the meanwhile,the control software and program interface were also designed.Secondly,the Camshift algorithm is adopted as the target tracking method.By analyzing the application environment and the shortness of Camshift,the improvement method of Camshift algorithm was proposed.Firstly,using Hue and texture to construct feature histogram,which aims at solving the problem of single characteristics not being able to described target appearance completely.Secondly,in order to enhance the discrimination of background and target,the target is weighted by local background.Thirdly,due to the incompetence of Camshift to detect fast moving target tracking,the Kalman filter is used to predict the initial value of the search window.Hence,the experiment results were shown the robust of complex background and fast moving objects by an improving algorithm.Thirdly,in order to expand the range of camera view,The PTZ control mechanism was installed in the vision tracking module.The Kinect's internal parameter matrix was calibrated by using Zhang Zhengyou's method.The target coordinate was calculated by according to thecamera geometry and target depth values,The putting forward of a PTZ control method is based on kinematics characteristic of the mobile robot and target position.Thus,The experimental result shows this method has a suitable stability and accuracy.Finally,in order to realize the obstacle avoidance function of the mobile robot on 2-D and3-D ground,a bulge and pit detection method was proposed,which uses two laser radars,and an improved VFH+ obstacle avoidance algorithm was used to select the driving path.Experiment shows that the proposed method can effectively help mobile robot to avoid convex and concave obstacles and thus the obstacle avoidance function of the motile robot on the 2-D and 3-D ground is successfully realized.