Research On Indoor Pedestrian Tracking Mobile Robot System Based On ROS

With the development of science and technology and the gradual rise of human cost,mobile robots are increasingly widely used in the field of security and service.In this thesis,indoor monitoring equipment is combined with mobile robot to realize automatic tracking of pedestrians entering the monitoring area by mobile robot.Based on convolutional neural network,this thesis constructs the multiple cameras pedestrian monitoring tracking system,and proposes a new multiple cameras target matching method.At the same time,in the robot side,the mobile robot implements the positioning,obstacle avoidance,and other functions,and improves the local path planning algorithm for mobile robot to ease going curve phenomenon in the process of the mobile robot moving.The communication module based on Socket is developed to realize the concurrent communication between the monitoring system and the mobile robot.The work of this thesis mainly includes three parts:(1)Multi-camera pedestrian monitoring and tracking system.A camera was set up at a high place in the room to shoot indoor scenes for pedestrian tracking.By using the Deep SORT model,the target detector was modified from Faster R-CNN to YOLOv3,and the real-time and accuracy of target tracking were better.In order to expand the scope of monitoring,multiple cameras can be built.When the same target appears in multiple cameras,it needs to match the target in the picture of multiple cameras.In this thesis,the matching based on distance is used,and the matching logic is improved to match the target in two cameras better.(2)Mobile robot system.Based on Robot Operating System(ROS),a robot that can move autonomously and avoid obstacles is built.Mobile robot is mainly built using laser radar combining odometer information to realize localization and mapping.The communication module receives the pedestrian's location from the monitoring system,and then plans the path to the pedestrian's position.The mobile robot can bypass the obstacles when there are obstacles and achieve pedestrians tracking finally.In this thesis,the local path planning algorithm used by the mobile robot is improved to make the forward trajectory of the mobile robot close to a straight line.(3)Mobile robot pedestrian tracking system.Based on Socket programming,the concurrent communication module of monitoring tracking system and mobile robot system is developed to realize the communication between Windows and Linux platform under the same wireless LAN.The communication module will send the pedestrian position detected by the monitoring and tracking system to the mobile robot as a target.The robot can move to the target point.Meanwhile,the robot will also send its position to the monitoring and tracking system through the communication module.The joint debugging and integration of monitoring system and mobile robot has realized the function of tracking indoor pedestrians.