Research On The Technology Of Body Size,Position And Posture Recognition Based On Point Cloud In Loading System

As one of the important links of transportation automation,the material packaging vehicle link has received extensive attention.At present,the corresponding material packaging vehicle system has been put into use on the market,but there is a general lack of research on the automatic identification of the size,position and orientation of the car body.The size,position and orientation of the car body need to be measured manually,with low efficiency and poor accuracy,and the loading system is difficult to achieve automation.In this paper,a car body size,position and orientation recognition system based on laser radar point cloud is developed.The filtering processing of point cloud data and the research on the car body size,position and orientation recognition method are focused on.The car body size,position and orientation recognition is completed,which provides a basis for the automation of the loading system.The main research contents and achievements of the paper include:1.Design overall scheme of bucket size pose recognition system.The demand of the loading system for the automatic recognition of the size and pose of the car body is analyzed.A car body size and pose recognition system based on lidar point cloud is constructed,and the mechanical structure and software structure of the system are designed.Analyze and compare the accuracy and stability of the equipment to determine the selection of laser radar and rotating platform;the realization technology of Modbus and UDP communication based on MATLAB is introduced.2.Point cloud filtering based on noise classification.According to the data collected by the laser radar and the angle of the rotating platform,a three-dimensional point cloud is generated.Starting from the analysis of the causes of noise in the point cloud,combined with the characteristics of the loading system and the working environment,the noise is divided into scene noise and measurement error noise.For the scene noise,according to the installation position of the system,the direct filtering combined with statistical filtering is used for filtering.Aiming at the measurement error noise,according to the position relationship between the normal vector and the coordinate system of the bucket point cloud,an algorithm for filtering according to the angle between the normal vector and the coordinate axis is proposed.3.A bucket size pose recognition method based on point cloud slicing is proposed.On the basis of traditional point cloud slicing,by calculating the average distance correlation coefficient of point cloud,the thickness of point cloud slices is directly related to the number,reducing the number of point clouds to be processed by a single slice.Through the point cloud slice processing of different datum planes,the size and position information such as the length,width,height and pose of the bucket are obtained.On the basis of the height of the bucket baffle obtained by the point cloud slicing process based on the vertical Y axis,a method of removing the point cloud of the locomotive based on the height difference between the locomotive and the bucket is proposed,which realizes the removal of the point cloud of the locomotive.4.Experimental platform construction and experimental verification.A set of experimental platform for the detection of the size and pose of the vehicle is built,the motion control software of the system is compiled,and the man-machine interface of the recognition system of the size and pose of the vehicle is developed.The experimental results show that the bucket size pose recognition system can work effectively in the laboratory and outdoor environment,and the bucket size pose recognition speed is fast and the accuracy is high.