3D Assembly Verification Using AI Point Cloud Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for checking the correctness of complex assemblies, such as in machinery and plant engineering, are time-consuming and computationally intensive due to the need for multiple perspective views, and are limited by the use of 2D or 2.5D projections that do not fully capture the 3D reality.
Innovation Solution
A computer-implemented method that determines an actual 3D model of the assembly from image data and compares it with a target 3D model to check correctness, using a single model for comprehensive verification from multiple perspectives, employing artificial intelligence to identify components and determine their positions and orientations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple perspective views are used to check complex assemblies, then measurement precision and completeness of verification are improved, but device complexity and computational intensity increase
Solution Approach 1:
The patent transitions from 2D image projections to 3D point cloud models and 3D convolutional neural networks. By representing assembly components in three-dimensional space with point clouds, the system achieves comprehensive verification from multiple perspectives simultaneously without requiring multiple separate 2D views, thus improving measurement precision while managing device complexity through unified 3D processing
Solution Approach 2:
The patent employs a universal 3D convolutional neural network that can process various types of assembly verification tasks (component presence, position, orientation, alignment) through a single unified model. This multi-functional approach eliminates the need for separate processing pipelines for different verification aspects, reducing device complexity while maintaining comprehensive verification accuracy
2Measurement precision
If multiple perspective views are used to check complex assemblies, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The patent merges multiple perspective views into a single unified 3D point cloud model. Instead of processing multiple 2D images separately and then integrating results, the system captures all spatial information in one 3D representation and performs verification in a single unified process, significantly reducing inspection time while maintaining the measurement precision benefits of multi-perspective verification
Solution Approach 2:
The patent performs preliminary 3D model generation and component identification before detailed verification tasks. By pre-processing the image data into a structured 3D point cloud model with identified components and their spatial relationships, the system prepares the data in advance for various verification operations, enabling faster execution of specific checks without reprocessing raw images
3Device complexity
If 2D or 2.5D projections are used for assembly verification, then device complexity is reduced, but loss of information occurs
Solution Approach 1:
The patent explicitly addresses information loss in 2D projections by transitioning to 3D point cloud representations. The point cloud model preserves complete spatial coordinates (x, y, z) for all visible points on components, maintaining full spatial information including depth, orientation, and relative positioning that is inherently lost when projecting 3D objects onto 2D planes
Solution Approach 2:
The patent creates a digital 3D copy of the physical assembly through point cloud modeling. This virtual 3D replica accurately reproduces the spatial structure, component positions, and geometric relationships of the actual assembly, preserving all spatial information in a format that can be processed and analyzed without the information degradation that occurs with 2D projections
Data Source
AI summary
A computer-implemented method configured to check a correctness of an assembly is provided, wherein the method comprises determining an actual 3D model of the assembly based on image data relating to the assembly, and comparing the determined actual 3D model with a target 3D model of the assembly to check the correctness of the assembly.