Research Of Detecting Method For Baggage Type Of Aircraft Cargo

At present,the loading and unloading links of air cargo transportation mostly depend on manual operation,which will lead to some problems including human injury and low work efficiency.For improving the automation level of cargo loading and unloading,this paper investigates the detection of stack type in automatic palletizing system.This investigation aims to reduce the impact of problems including cargo slippage and deformation on the palletizing planning process during palletizing.Based on the investigation of the key technologies including point cloud topology construction,scattered point cloud boundary feature extraction and stacked point clouds corner detection,a method for detecting stacked point clouds corner is designed and implemented.To solve the problem of fast searching near-neighbor point of scattered point cloud under a large amount of data,the process of establishing topological relation of scattered point cloud space by spatial cells,octrees and KD trees are analyzed.The three-dimensional KD tree is used to establish the spatial topology of scattered point clouds,and then the neighbor points can be found quickly by using the neighbor point search algorithm.Designing a feature extraction method suitable for the detection of the boundary points of stacked point clouds.By analyzing and comparing common boundary point identification methods,the method of combining density estimation and field force method is used in this paper to extract the boundary points of stacked point clouds.First of all,the nearest neighbor of the sampling point was found by KD tree;next,the distance from the sampling point to the mode point and the resultant force between the point and the nearest neighbor point are calculated;and then,the point larger than the threshold is defined as the boundary point by using the criterion of boundary point determination;finally,the boundary point identification experiment is conducted using simple stacked point clouds data.Investigating the corner detection method for multi-level stacked point clouds.For the problem of corner points detection of stacked point clouds,on the premise of the obtaining boundary points of the stacked points,this paper analyzes the angular characteristics of the stack point corners in the plane point clouds,and proposes a method for stacking angle detection based on point clouds clustering.Firstly,the candidate corner points are obtained by using the improved sharpness corner detection algorithm;and then,the improved fast density peak search algorithm combined with K-means algorithm is used to complete the clustering of candidate corner points;finally,the candidate corner points were merged and calculated by density value method,the results are the corner coordinates of stack type.Based on the above investigation,the experiment for stack type detection is carried out by using Kinect to collect the 3D point clouds data of complex carton stack type.The experimental results indicate that on the premise of obtaining the boundary points of multi-layer and non-closed planar stacked point clouds,the corner coordinates of different depth in stacked point clouds can be calculated by using the angle features of local regions and the clustering of point clouds;the error of corner coordinates is controlled within 8 mm.