An Improved Supervoxel Clustering Algorithm of 3D Point Clouds for the Localization of Industrial Robots

Abstract

Supervoxels have a widespread application of instance segmentation on account of the merit of providing a highly approximate representation with fewer data. However, low accuracy, mainly caused by point cloud adhesion in the localization of industrial robots, is a crucial issue. An improved bottom-up clustering method based on supervoxels was proposed for better accuracy. Firstly, point cloud data were preprocessed to eliminate the noise points and background. Then, improved supervoxel over-segmentation with moving least squares (MLS) surface fitting was employed to segment the point clouds of workpieces into supervoxel clusters. Every supervoxel cluster can be refined by MLS surface fitting, which reduces the occurrence that over-segmentation divides the point clouds of two objects into a patch. Additionally, an adaptive merging algorithm based on fusion features and convexity judgment was proposed to accomplish the clustering of the individual workpiece. An experimental platform was set up to verify the proposed method. The experimental results showed that the recognition accuracy and the recognition rate in three different kinds of workpieces were all over 0.980 and 0.935, respectively. Combined with the sample consensus initial alignment (SAC-IA) coarse registration and iterative closest point (ICP) fine registration, the coarse-to-fine strategy was adopted to obtain the location of the segmented workpieces in the experiments. The experimental results demonstrate that the proposed clustering algorithm can accomplish the localization of industrial robots with higher accuracy and lower registration time.

Fast and High Accuracy 3D Point Cloud Registration for Automatic Reconstruction From Laser Scanning Data

“…However, the ICP algorithm suffers from issues such as low computational efficiency, the tendency to fall into the local optimum, and high dependence on the initial poses. To overcome these challenges, several improvement algorithms have been proposed, such as GICP [8], NICP [9], and SACICP [10], [11]. Among these, SACICP divides the original ICP into two steps: coarse registration and fine registration.…”

“…Among these, SACICP divides the original ICP into two steps: coarse registration and fine registration. Firstly, the FPFH of the point cloud is computed by the Sample Consensus Initial Registration (SAC-IA) algorithm [11], which finds the corresponding points with similar FPFH features in the target point cloud and calculates the rigid body transformation matrix between the corresponding points to achieve coarse registration. The rough registration step of the SACICP algorithm partially addresses the computational efficiency and initial registration dependence issues of the ICP algorithm.…”

“…Let X and Y be the multi-dimensional vectors of the source point cloud interval four-point set and the corresponding target point cloud interval four-point set, respectively. The feature similarity coefficient r XY of X, Y can be calculated using equation (11).…”

3D object recognition for coordination-less bin-picking automation

“…This is because we believe that deep learning methods can construct recognition models with better generalization performance than the inflexible algorithm-based methods which have been researched in the past [3]. Furthermore, since algorithm-based methods have a problem with execution time, we research a method based on deep learning in this study [4].…”

Automatic robotic trajectory planning assisted by laser projection measurement

“…Trajectory planning is a key to reliable microscopic inspection. The trajectory planning based on the point clouds is currently developing rapidly [5][6] . Laser projection imaging is a convenient technique to acquire point clouds from complex components, but it applies to diffuse objects only.…”