Research On Obstacle Detection And Leader Recognition Method Based On Lidar For Mobile Robot

Obstacle detection and leader recognition are two important issues that need to be solved in mobile robot environment perception.3D Lidar has strong environmental adaptability and can directly obtain more accurate distance information.It is a commonly used sensor for mobile robots.In this thesis,for obstacle detection and leader identification methods,four aspects had been studied: lidar perception system scheme construction,perception system calibration and data preprocessing,static obstacle detection,dynamic obstacle grid removal and leader recognition.Through the analysis of the specific requirements of the mobile robot's environmental perception,sensing system scheme of using 32-line and 16-line lidars in combination was designed while taking into account the maximum performance of the sensor,and a set of dual-lidar sensing system equipment was processed and assembled.The parameter calibration method between lidars and between lidar and rack based on RANSAC was studied.Calibration experiments were carried out to verify the correctness of the above calibration method,and the numerical value of each parameter in statistical significance was obtained.In addition,distortion removal of point cloud and time synchronization between lidars were studied for the sensing system,and experiments had proved the effectiveness of the above method.The processing flow of static obstacle detection was formulated.Based on the processing flow,a real-time grid map containing obstacles and passable areas was established during the experiment,and the functions of multi-frame point cloud matching and multi-frame grid map splicing were realized,which verified the correctness of the static obstacle detection theory.A dynamic obstacle grid removal method based on investigating the neighborhood of the obstacle grid in front frame was proposed and verified by experiments.This method was used to suppress the trailing phenomenon caused by dynamic obstacles during grid splicing.The leader identification method based on a high reflectivity tag was studied,and the effect of simultaneous leader identification and obstacle detection had been realized through experiments.