Virtual 3D Dataset Generation for Bulk Workpiece Pick Positions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Generating a large number of precise learning datasets for training machine learning models that estimate take-out positions for bulk-loaded workpieces is time-consuming and labor-intensive, as it requires frequent reconfiguration of actual workpieces in different forms within a container.
Innovation Solution
A learning dataset generation device and method that utilize three-dimensional CAD data to generate virtual imaging objects of workpieces in different forms within a container, acquire virtual distance images using a virtual three-dimensional measurement machine, and associate teaching positions with these images to create a large number of learning datasets efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If actual workpieces are reconfigured in different forms within a container to generate diverse learning datasets, then the quantity and diversity of learning datasets improve, but the time and labor required for dataset generation increase significantly
Solution Approach 1:
The patent creates virtual copies of workpieces and containers in a three-dimensional virtual space using CAD data. Instead of physically reconfiguring actual workpieces multiple times, the system generates multiple virtual imaging objects by digitally arranging workpiece models in various configurations within virtual container models, thereby obtaining diverse distance images without physical manipulation.
Solution Approach 2:
The patent replaces the mechanical system of physically handling and reconfiguring actual workpieces with a virtual measurement system. A virtual three-dimensional measurement machine performs measurements on virtual imaging objects in computer-generated environments, eliminating the need for repeated physical setup and measurement operations.
2Adaptability or versatility
If actual workpieces are reconfigured in different forms within a container to generate diverse learning datasets, then the diversity of learning datasets improves, but the labor intensity and operational burden increase
Solution Approach 1:
The system uses digital copies (CAD models) of workpieces and containers to create diverse scenarios. By manipulating virtual objects rather than physical ones, the system achieves high adaptability in generating varied learning datasets while eliminating the manual labor of physical reconfiguration.
Solution Approach 2:
The system achieves diversity by changing parameters such as the number of workpieces, their positions, orientations, and container configurations in the virtual environment. This allows rapid generation of diverse learning scenarios through parameter adjustment rather than physical manipulation.
3Measurement precision
If a three-dimensional measurement machine is used to acquire distance images of actual bulk-loaded workpieces, then measurement precision is maintained, but the process becomes time-consuming and labor-intensive when repeated for different configurations
Solution Approach 1:
The patent replaces the physical three-dimensional measurement machine with a virtual measurement machine that operates in computer-generated space. The virtual measurement machine calculates distance images from three-dimensional coordinates of virtual imaging objects, maintaining measurement precision while enabling rapid generation of multiple datasets without physical setup time.
Solution Approach 2:
The system performs preliminary actions by pre-defining the three-dimensional models of workpieces and containers, and pre-establishing the virtual measurement machine's measurement capabilities. This allows rapid generation of distance images for multiple configurations without repeated physical measurement setup.
Data Source
AI summary
A learning dataset generation device includes: a memory that stores three-dimensional CAD data of a workpiece and a container; and one or more processors including hardware, wherein the one or more processors are configured to use the three-dimensional CAD data of the workpiece and the container, stored in the memory, to generate, in a three-dimensional virtual space, a plurality of imaging objects in which a plurality of the workpieces are bulk-loaded in different forms inside the container, acquire a plurality of virtual distance images by measuring each of the generated imaging objects by means of a virtual three-dimensional measurement machine disposed in the three-dimensional virtual space, accept at least one teaching position for each of the acquired virtual distance images, and generate a learning dataset by associating the accepted teaching position with each of the virtual distance images.