Research On Outline Dimension Inspection Method Of Cargo Vehicle Based On UAV LiDAR

Overloading and over-limited transportation of freight vehicles have seriously damaged highway infrastructure,causing damage to roads and bridge fractures.The existing vehicle size detection technology has the disadvantages of fixed detection location and high cost,which makes it difficult to achieve real-time dynamic detection of overloading and over-limited behavior of vehicles.The UAV platform has the advantages of low cost,high flexibility,and highly real-time,so this paper introduces the UAV platform,collects vehicle point cloud data by carrying LiDAR,and constructs a vehicle outline size detection method based on various point cloud processing algorithms to realize the portable and mobile detection of vehicle outline size.The specific research contents are as follows.(1)A vehicle outline dimension measurement system based on UAV LiDAR was built.The quadrotor UAV,which is widely used nowadays,is used as the flying platform,and the whole body of the UAV is made of carbon fiber material to improve the UAV range.In terms of measurement equipment,the working principles,advantages,and disadvantages of different types of LiDAR were analyzed in depth,and the mechanical LiDAR suitable for this study was selected.(2)A real-time wireless data transmission module between the UAV LiDAR and the ground-based host computer is designed.The vehicle outline dimension measurement system based on UAV LiDAR has very high requirements for both uplink and download rates of data transmission.To ensure that the measurement system can carry out real-time data transmission stably,this study obtained the router development board by disassembling and replacing the router development board chip and Wi Fi module to meet the test requirements.To ensure the stability of data transmission,a Wi Fi signal amplifier is added to the upper computer.The designed real-time wireless data transmission module has been tested to support the measurement system to maintain normal wireless communication with the host computer within a range of 30 m.(3)The original point cloud data format conversion was completed and the point cloud pre-processing algorithm was designed.Firstly,the point cloud data format is converted by calling the relevant commands of ROS in the Ubuntu system.Then,based on the Point Cloud Library algorithm library,the pre-processing process of the original point cloud data is designed and completed,mainly consisting of point cloud topology relationship establishment,point cloud filtering,and point cloud clustering algorithm.The noise points and outlier points in the vehicle point cloud data are removed to obtain the complete point cloud data of the vehicle under test,which provides an effective input data set for the subsequent vehicle outline dimension measurement.(4)A vehicle dimension detection method based on Principal Component Analysis to construct a minimum wrap-around box is designed and completed for actual vehicle testing.The results of the experiments on the currently used point cloud enclosing box algorithms show that several algorithms are unable to obtain accurate vehicle outline dimension measurement results.Therefore,we designed an algorithm based on PCA to construct the minimum enclosing box of the point cloud to achieve the accurate measurement of vehicle outline dimensions,and the measurement error is within 4.0 %after the test results of three vehicle models.