Research On Environmental Perception Algorithm Of Low Speed Unmanned Logistics Vehicle Based On Lidar

With the rapid development of the domestic logistics industry,the importance of the "last mile" terminal distribution service in the logistics industry continues to increase,and the problems of high distribution costs and low efficiency are increasingly prominent.Lowspeed unmanned logistics vehicles commonly used in specific scenarios such as parks,communities,and campuses have gradually become the focus of attention of major logistics companies and related research institutions.The role of the unmanned vehicle environment perception system is just as important as human eyes.Its main environment perception content includes the detection of road boundaries and the tracking of road obstacles.Based on the cost constraints and perception capabilities of the low-speed unmanned logistics vehicle perception system,this paper selects a relatively low-cost single-line lidar to develop a Labview environmental information collection system with good visibility and real-time performance.Through the actual vehicle experiment,the accuracy and effectiveness of the road boundary detection algorithm,obstacle clustering algorithm and obstacle target tracking algorithm under low speed conditions are verified,which has certain academic research value and engineering value.The main research contents of this article are as follows:(1)Introduces the working principle and performance of single-line lidar,and the basic structure of Labview-based environmental information collection system.Write the radar point cloud information analysis program in Labview to analyze the lidar message information.The Labview environmental information collection system calls a dynamic link library DLL that includes road detection algorithms,obstacle clustering,and tracking algorithms to complete the collection of environmental information.(2)A detection algorithm for road passable area of lidar is proposed.According to the characteristics of the point cloud scanned by the two-dimensional lidar,the nearest neighbor algorithm is used to remove noise.Remove the road surface part of the inverted “U” type data of the radar point cloud,extract the corner points of the remaining inverted “L” type cloud,obtain the roadside points through slope control,and finally use the least square method to determine the specific location of the road boundary.The determination of the road boundary provides security for the driving of low-speed unmanned logistics vehicles.(3)Aiming at the problem of obstacle clustering and tracking,an adaptive threshold clustering method,multi-parameter correlation and target tracker management rules are proposed.According to the point cloud shape of the road obstacles scanned by the lidar,the region of interest is set,and the improved DBSCAN clustering algorithm is adopted.Considering that the farther the point cloud is,the more sparse the point cloud is.An adaptive dynamic threshold is adopted to ensure the integrity of the point cloud information of the same obstacle.The multi-parameter correlation algorithm based on the nearest neighbor algorithm is used to determine whether the point cloud information at different times is the same obstacle by solving the judgment function including the displacement,length,width,speed and direction of the obstacle target.For the loss of vehicle information during driving,the Kalman filter algorithm is used to predict the key information such as the position and speed of obstacles at the next moment,which is provided to the unmanned vehicle decision system to ensure the safety of driving.By setting the target tracking manager,the existence threshold and the loss threshold of obstacles are used to ensure the reliability of the tracked targets,avoid the possibility of false alarms,and ensure the efficiency of target tracking.(4)Carry out the actual vehicle test and experimental verification by using the low-speed unmanned vehicle as the experimental vehicle.The results show that the research method in this paper can better detect the boundary information of the road ahead,has a good clustering effect of obstacles,and can effectively track dynamic obstacle targets with a speed of 1-20 km/h.