Design And Implementation Of ROS Mobile Robot Tracking System

With the rapid development of robot and artificial intelligence technology,service robots have also been developed rapidly,people's demand for service robots is increasing.The tracking function is an important content of the research in the field of service robots.The robot operating system(ROS)has a distributed open source software architecture that abstracts the underlying hardware of the robot and improves the development efficiency of the code,which has attracted the attention and research of many scholars and developers.In the thesis,the ROS system is used to design the target tracking and face tracking of mobile robot.It can automatically adjust the tracking speed and maintain a safe tracking distance.In this paper,the hardware platform of the mobile robot is built,Kinect2.0 vision sensor,the wireless network card intel 7260 AC HMW drive system for the ROS system and the Kobuki software system are configured.The Ubuntu14.04 and ROS indigo operating systems are set up on the lower computer,and an automatically connected wifi is set up in the system.The environment variables of the.bashrc file are set to facilitate the development and testing of mobile robot.The main contents of the thesis are as follows:1.MeanShift tracking node,Camshift tracking node and Kalman tracking node are designed under ROS framework.The design process of three tracking nodes is elaborated.The advantages and disadvantages of each tracking node are analyzed through experiments.2.Aiming at the problem of poor real-time tracking of mobile robot,the real-time target tracking node is designed.According to the prior information of the target,Kalman filtering is used to predict the target position,and Camshift algorithm is used to search the target based on the prediction information,which improves the processing speed of the system,and shortens the time of target search,meets the real-time requirement of the system.For the complete occlusion problem of target tracking,the state equation and observation equation of the target are established,and the occlusion tracking node is designed.According to the target state information before occlusion,the target is predicted and tracked.After the occlusion is over,the target is re-locked,and automatically switched to the real-time tracking mode.Experimental results verify the real-time and robustness of the tracking system.3.The face tracking node uses face detection node,sub-pixel corner detection node and optical flow tracking node to realize the face tracking of the mobile robot.The face detection node in the function block can accurately detect the face in the image.For the tracking delay problem of the face detection node and the image display delay problem,the tracking strategy is changed and the feature points of the face area are tracked.The corner detection node is designed to extract the feature points of the human face area,and then the optical flow tracking node is used to track these feature points,realizing the real-time tracking of the face of the mobile robot.For the problems of high computational complexity of the face tracking node and the color interference of the target tracking,a color interference target tracking node is designed to solve the real-time and color interference problems of the mobile robot face tracking under certain conditions.Thus the real-time requirement of face tracking is satisfied,and the tracking efficiency of the system is improved.Aiming at the occlusion of face tracking,the face occlusion tracking node is designed to solve the occlusion problem in the face tracking process,improve the robustness of the tracking system,and better realize the face tracking function of the mobile robot.