Research On Obstacle Detection Method Based On Lidar In The Driving Area Of Smart Car

With the automobile becoming more and more intelligent,the key technology of intelligent vehicle is considered as the revolutionary technology of automobile industry.In this paper,vlp-16 line lidar,which has the advantages of moderate amount of data,strong anti-jamming ability and high measurement accuracy,is used to study the obstacle detection technology in the driving area of intelligent vehicle.The intelligent vehicle senses the surrounding environment under the structured road,and provides environmental information for vehicle path planning,active braking,active obstacle avoidance,etc.Firstly,in order to obtain the accurate LIDAR point cloud,the LIDAR point cloud is calibrated with the data information obtained from GPS / IMU.In order to extract the LIDAR point cloud up and down,the laser scanning lines are reordered.On this basis,three coordinate systems are established,the calibrated laser point cloud is generated in the lidar coordinate system,and the laser point cloud is transformed from the lidar coordinate system to the vehicle body coordinate system for unification,and then transformed to the image coordinate system to display the detection results.Then,the boundary extraction process of driving area is introduced.In the first step,a multi-attribute grid map is established to store the extracted environmental information;in the second step,in order to better and faster segment the ground point cloud,a multi-range iterative plane fitting method is proposed to segment the point cloud of lidar calibration,and obtain the ground point cloud and non ground point cloud;in the third step,in order to avoid the false detection and missing detection that are easy to occur during the extraction of road edge,a new method is proposed In this paper,a multi feature layered method is proposed to extract accurate and reliable road boundary candidate points for each laser scanning layer.Then,an obstacle detection method based on grid neighborhood relationship is introduced.The obstacle clustering algorithm is analyzed,and the data is processed by region condition filtering and grid filtering.When processing LIDAR point cloud data,a clustering algorithm based on grid neighborhood is proposed according to the point cloud distribution in grid map.Combined with Euclidean distance clustering,it can not only find clusters of arbitrary shape,but also reduce the amount of data processed.In this algorithm,the occupied grid is represented as coordinate points,and it doesn’t pay attention to the specific point attributes in the occupied grid.This processing methodcan improve the real-time performance of the system and the accuracy of clustering.Finally,a large number of real-time experiments are carried out on the intelligent vehicle platform of our laboratory to verify the real-time and accuracy of the proposed methods of boundary extraction and obstacle detection,and the experimental results are analyzed and discussed.