Research On Key Technologies Of Rapid Surveying Method For Terrain In Front Of The Vehicle Based On Lidar

Lidar is a new type of sensor that can quickly acquire high-density,high-precision three-dimensional coordinates within a scanning range without touching the measured object.It is a practical and efficient measurement method and technology for modeling roads and terrains.The information measurement field has been widely used.This article combines the national key R&D project “Study on key chassis and suspension technologies for high-mobility emergency rescue vehicles(including fire vehicles)”(Project Number: 2016YFC0802902).In order to achieve the purpose of developing a special suspension system,the front of the vehicle scanned by the laser radar is obtained.Topographic point cloud data is studied.After clustering analysis,altitude information extraction,rasterization and other processing of terrain point cloud data,a terrain map grid with high information is updated in real time,and key algorithms are validated by setting up a corresponding development environment.The specific work is as follows:First,introduce the front terrain rapid measurement system.The composition of the laser radar is described in the form of a block diagram,and the principle of the distance measurement of the laser radar is known.Refer to the user manual of the model of the laser radar and master the characteristics and working parameters of the laser radar.The role of other parts of the terrain measurement system was introduced.Finally,the information processing method was illustrated in the form of a flow chart.Second,in-depth understanding of point cloud data.Through the calibration,the reference coordinate system of the point cloud data is transformed into the car coordinate system,and then the translation,rotation and scaling of the point cloud coordinate in the space coordinate system are realized by the matrix operation.The concept of neighborhood and K-D tree spatial index data structure are introduced to achieve efficient spatial retrieval of point cloud data.Based on this,a statistical filtering algorithm and a point cloud smoothing algorithm are designed and verified with the PCL development environment.Again,clustering analysis of point cloud data.The clustering algorithm classifies the discrete point cloud data according to its own similarity difference.According to the characteristics of the laser radar point cloud data,DBSCAN clustering algorithm is used for cluster analysis,and the algorithm is improved by the theory of nuclear density estimation.The least squares method is used to do boundary-matching processing on the clustered obstacles to distinguish between obstacles and non-obstacles.Then,information is extracted to obtain the height information of the entire terrain.Finally,a terrain information map is created using a rasterization method.Based on the evidence theory and terrain height information,each grid cell of the map is judged by state,and each grid cell is filled with data using the maximum value to obtain a rasterized topographic map containing the maximum height information.Considering that the car is in motion,a grid cell registration method based on the angle and speed of the car movement is adopted,and the global map of the previous frame is dynamically updated by using a frame of local map information,and a global map updated in real time is finally obtained.