8 results about "Plane segmentation" patented technology

This invention discloses a method and system for lane line vectorization pre-annotation based on multi-frame LiDAR point clouds. It constructs a global dense point cloud map by performing spatial coarse filtering and spatiotemporal synchronous stitching on the original point cloud data; extracts the main road surface point cloud through elevation fine filtering and plane segmentation; suppresses and enhances non-uniform backgrounds based on morphological top-hat transform; extracts candidate lane line point clouds using adaptive threshold segmentation combining statistical lower bound and recursive maximum inter-class variance method; extracts structured candidate lane lines and geometric features based on weighted iterative RANSAC line fitting guided by both intensity and longitudinal density; and constructs a global topology graph model and a multi-dimensional scoring mechanism to screen and resolve conflicts among candidate lane lines, outputting the final lane line vector map. This invention effectively overcomes uneven intensity and local noise interference, achieving high-purity and high-robustness lane line vectorization reconstruction, and significantly reducing the cost of manual annotation.

The application discloses a kind of applied to the three-dimensional model retrieval method, system, equipment, medium and product of symmetrical depth image, it is related to three-dimensional model retrieval technical field, obtain multiple standard three-dimensional models, respectively to each standard three-dimensional model is carried out symmetrical plane segmentation processing, obtain multiple target symmetrical planes, according to each target symmetrical plane and pre-trained three-dimensional feature extraction network respectively to each standard three-dimensional model is carried out three-dimensional feature extraction, obtain corresponding three-dimensional feature library, according to each target symmetrical plane and pre-trained three-dimensional feature extraction network respectively to each standard three-dimensional model is carried out three-dimensional feature extraction, obtain corresponding three-dimensional feature library. Overcome the three-dimensional model retrieval method based on two-dimensional image at present multiple dependence multi-view image or depth image carries out feature learning, there is feature extraction process complex, retrieval performance depends on network pretraining effect and so on problem, reduce the technical problem of three-dimensional model's retrieval performance and efficiency.

This invention provides a robot painting path planning method based on deep proximity clustering, belonging to the field of machine vision and automated robot painting technology. The method includes the following steps: First, obtain a triangular mesh model of the workpiece to be painted and perform geometric plane segmentation on the triangular mesh model; second, discretize the robot's movement space, and aggregate reachable planes into candidate continuous working ranges based on deep proximity clustering to ensure motion continuity during a single stop; then, model task selection as a set coverage problem to solve for the minimum number of stops, and model the execution order problem of tasks within the selected continuous working range as a traveling salesman problem to optimize the execution order; finally, through a two-layer global optimization of set coverage and the traveling salesman problem, generate smooth trajectory code. This invention solves the problems of low programming efficiency, long robot idle travel, and poor coating uniformity in existing large and complex workpiece painting processes, and is suitable for fully automated, high-efficiency, and high-quality painting operations on large and complex workpieces such as marine steel structures.

The application discloses a kind of spraying robot process parameter driving path planning method and system, it is related to the field of spraying robot spraying path planning.The method will process parameters as the constraint condition and driving input of path generation and motion control, then through geometric modeling and rule constraint makes process parameters directly participate in path generation, combined with the double semantic driving dynamic adjustment path point density and step parameter of "surface-thickness", then through intelligent path planning completes spraying plane segmentation, path point generation, obstacle avoidance planning and trajectory optimization, generates optimal spraying path, finally converts spraying path into motion instruction, combined with perception feedback realizes the dynamic closed loop optimization of spraying path.The application realizes the deep coupling of spraying process and robot motion planning, process parameters directly drive path generation, without manual demonstration, spraying path can adapt to workpiece morphology and process requirement, ensure the consistency and stability of spraying quality, improve spraying operation efficiency.

The present application belongs to the field of computer vision, and particularly relates to a bolt defect detection method based on image information and point cloud information, comprising: collecting an inspection image and point cloud data of an inspection area containing bolts to be inspected, wherein the inspection image and the point cloud data are pixel-aligned; based on the inspection image, a pre-trained target recognition model is used to regress and predict a target frame of the bolt to be inspected; if the predicted result is that there is no bolt to be inspected, the bolt defect detection is completed; if the predicted result is that there is a bolt to be inspected, point cloud mapping of the bolt area is performed according to the target frame obtained by regression, and a RANSAC plane segmentation algorithm is used to segment and extract the upper surface of the bolt and the mounting reference surface in the point cloud of the bolt area; the distance between the two planes is calculated and compared with a set threshold value, so as to determine whether the bolt is loose, and the bolt defect detection is completed. The present application can efficiently realize bolt defect detection.

The present application belongs to the field of industrial automation and robot welding technology, and specifically relates to a welding seam feature extraction and intelligent planning method based on multi-view fusion, which comprises the following steps: 1) acquiring point cloud data of a steel member under multiple views through a camera arranged at multiple photographing positions; 2) pre-processing the point cloud data and performing multi-plane segmentation to segment multiple different plane point cloud data and extract a target plane; 3) sequentially calculating the intersection straight line of all adjacent target planes and the starting point position and posture of the corresponding straight line to obtain all straight line welding seam track information under the point cloud; and fusing the straight line welding seam track under each view to obtain the fused welding seam track; 4) planning an optimal moving path of a robot welding gun; and 5) performing secondary optimization to obtain complete high-precision steel structure welding seam track information. The present application has strong adaptability to light changes and surface reflections through the proposed multi-view scanning and track fusion technology.

The application provides a model generation method and device and electronic equipment, the method comprises the following steps: obtaining a to-be-processed model; a plurality of first planes perpendicular to a first direction axis are set on the to-be-processed model, the to-be-processed model is divided based on the plurality of first planes, and a first structure is obtained; a plurality of second planes perpendicular to a second direction axis are set on the to-be-processed model, the to-be-processed model is divided based on the plurality of second planes, and a second structure is obtained; and a target model with a waffle structure is generated based on the first structure and the second structure. After the to-be-processed model is obtained, the to-be-processed model can be automatically segmented to obtain a model with a waffle structure. Compared with a manual method of making a model with a waffle structure, the method automatically completes model making through a program, a large amount of manpower and time are saved, and the efficiency of model making is improved.

This application provides an adaptive control method and apparatus for a bin picking robot. The method includes: constructing a multimodal perception integration module to simultaneously collect environmental perception data and 3D point cloud and 2D image data of the bin; adaptively determining filtering parameters based on point cloud density and noise level, performing filtering and deep learning missing data completion, and then fusing the preprocessed point cloud with the image; quickly locating the region of interest using the 2D image, performing planar segmentation within the region and removing supporting planes, and using a curvature-constrained clustering algorithm to accurately segment the remaining point cloud to obtain independent point cloud clusters; extracting grasping features based on the independent point cloud clusters and generating action commands to complete the picking process. This application can improve the robot's adaptive capability and warehouse management efficiency.