Analysis And Registration Of On Board Lidar Point Cloud Data

Highly maneuvering emergency rescue vehicles are usually operated at the rescue site.The environment is harsh and the road conditions are complex,which seriously affects the ride comfort and speed of the vehicle.In this paper,the active suspension is used to mitigate the impact of the road on the vehicle,thus improving the ride comfort of the vehicle.The preview active suspension mainly senses the elevation information of the road surface in front of the vehicle through a sensor mounted on the vehicle,and then adjusts the working state of the suspension system in real time through the controller,thereby improving the ride comfort of the vehicle.Since the vehicle is in a high-speed driving state,the on-board lidar continuously collects point cloud data of the road surface in front of the vehicle,so the point cloud data coordinate system obtained at different times and at different angles of an object in front of the vehicle exists more or less.Deviation.To this end,this paper studies the precise registration method of lidar data.This project relies on the national key research and development plan "high-mobility emergency rescue vehicle(including fire truck)special chassis and suspension key technology research"(project number: 2016YFC0802902),using laser radar,IMU,GPS and computer equipment to build an experimental platform,The terrain in front of the vehicle is scanned,and the collected lidar point cloud data is preprocessed and registered.The paper mainly completed the following work:First,the front terrain scanning system is introduced and the point cloud data is preprocessed.Using laser radar,inertial measurement unit,GPS and on-board computer to build the front terrain scanning system to collect the distance,angle,vehicle attitude angle and position information of the front terrain in real time.Use the spatial grid method to establish the topological relationship of point cloud data.In the point cloud pre-processing process,the point cloud is compressed,the outliers and the point cloud smoothing are processed,and during the processing,the point cloud is also translated,scaled,and rotated to make it clearer and more clear.Comprehensive observation point cloud data.Secondly,point cloud feature description and feature point extraction are performed.The differential model of geometrical features of point cloud space is analyzed,and the normal and curvature are estimated and fitted.At the same time,the improved feature based on curvature extremum is used to extract feature points.Finally,the point cloud data is described using Point Feature Histograms.Finally,the traditional ICP point cloud registration algorithm is introduced and analyzed,and the advantages and disadvantages are summarized.A point cloud registration method based on the improved iterative closest point(ICP)algorithm is proposed,and three sets of experimental analysis are performed to compare the accuracy of the improved ICP algorithm.Sex and stability.