Research On Key Technologies Of Vehicle Frontographic Recurrence Based On Multi-Lidar

During the forward movement of the vehicle,the form of excitation and the undulating state of the ground have a great influence on the ride comfort.As an important part of the vehicle chassis,the suspension can alleviate the impact of uneven road surface on the vehicle body to a certain extent.How to sense the terrain and obstacle information in front of the vehicle and control the active suspension system to improve the ride comfort of the vehicle.It has attracted the attention of more and more researchers.This paper combines the national key research and development plan "Highmobility emergency rescue vehicles(including fire-fighting vehicles)special chassis and suspension key technology research"(project number: 2016YFC0802902),completes the data acquisition of the front terrain of the vehicle before the terrain terrain scanning perception system Further data processing work.The main work of the thesis is as follows:First,a data acquisition system based on multiple laser radars was designed and built.The hardware components(lidar,INS and GPS)of the front terrain scanning sensing system and its working principle are introduced,and the errors of the system platform are analyzed.The installation position and installation method of multi-laser radar are introduced,and the necessity of multi-laser installation is demonstrated and the multi-laser radar is jointly calibrated.Secondly,the point cloud data of the lidar is deeply understood.The importance and necessity of point cloud filtering and registration technology for vehicle terrain reconstruction are introduced.The related algorithms and accuracy evaluation methods of laser radar point cloud filtering and registration are described in detail.Various filtering algorithms and registration are analyzed and compared.The applicability,advantages and disadvantages of the algorithm and screening of filtering and registration algorithms suitable for this project.Again,the improved linear predictive filtering algorithm is used to improve the filtering effect.Considering that the traditional linear predictive filtering algorithm has better adaptability to different terrains,it has obvious advantages compared with other filtering algorithms,but it also has the disadvantage that the noise points have a great influence on the fitting of the late terrain trend surface.We improve the filtering algorithm and improve the efficiency and iterative termination condition judgment by noise point culling and data gridping block processing.The superiority of the improved linear predictive filtering algorithm is verified by comparative experiments.Finally,the precise registration of obstacles in front of the vehicle is accomplished using an optimized ICP registration algorithm.Because the ICP algorithm is easy to cause local optimal registration in the registration process,the efficiency and accuracy requirements for point cloud registration are getting higher and higher,and the traditional ICP algorithm is difficult to meet the requirements.In this paper,the principal component analysis method is introduced to improve the registration accuracy,and the kd-tree structure and the direction vector threshold are used to achieve accurate registration of 3D point cloud Finally,the effectiveness of the optimization algorithm is verified by experiments and finally the elevation map in front of the vehicle is obtained by surface fitting.