Design And Implementation Of Indoor Environment Mapping System Based On Millimeter Wave Radar

With the continuous rise of the Internet of Things and the continuous promotion of smart home applications in the home,users' acceptance of smart homes is getting higher and higher,and a series of smart product technologies represented by sweeping robots are developing faster and faster.Most of the sweeping robots on the market use lidar as the main sensor for building maps.In the same time,with the continuous improvement of FMCW millimeter-wave radar hardware and the continuous optimization of RF algorithms,its accuracy in distance and angle measurement has gradually improved,which also makes it possible to use millimeter-wave radar for mapping.The millimeter-wave radar mapping system starts from the principle input of millimeter-wave radar and improves the original process from the reflected wave processing stage.Then,an efficient millimeter-wave radar data acquisition module and a point cloud preprocessing module are designed.The interference point is filtered by nature,and then the octree is used to construct the occupied grid map.In order to offset the robot coordinate system caused by the error of the odometer,the posture correction technology,such as Monte Carlo-based particles The filtering algorithm makes the local point cloud more accurate when mapped to the global map,and obtains the initial global map.In terms of system optimization,the obtained global map is mainly subjected to morphological operations and cluster classification and other point cloud processing,and then the wall fitting is done for the part classified as the wall surface,so that the construction of the wall surface and obstacles is more accurate,To avoid problems such as wall dispersion and obstacle shape distortion.Finally,these results are visualized with ROS and the final results are presented on the RVIZ visualization tool.The entire system builds a node mechanism based on ROS.Each node uses multi-threads and message queues to communicate.The hardware uses TI's IWR6843-ISK for the radar and the Turtletle2 for the robot.Due to the characteristics of ROS node communication,the overall program includes nodes written in C ++ and Python,which makes C ++ for a large number of calculations that require performance,and some basic processing processes use Python to make the system architecture more clear and convenient.Integration of other modules such as navigation and path planning.