Position Posture Identification Using Virtual Measurement Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Three-dimensional measuring machines struggle to acquire complete point cloud data, especially for recessed areas, leading to increased processing loads and inaccurate feature matching, which hampers the identification of a target article's position posture, particularly in systems with limited resources.
Innovation Solution
A position posture identification device and method that includes actual measurement data acquisition, virtual measurement data generation, filter processing for affine transformation based on the measuring machine's properties, feature point extraction, and position posture calculation using coordinate transformation, which reduces unnecessary data and enhances accuracy by weighting feature points based on measurement probability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complete three-dimensional shape data is used for model data, then the processing load increases unnecessarily, but if only measurable parts are used, then feature matching accuracy improves
Solution Approach 1:
The patent extracts only the measurable parts of the target article from the complete three-dimensional shape data to create model data. This is achieved by identifying regions that the measuring machine can actually capture and excluding recessed areas or hidden surfaces that cannot be measured, thereby reducing processing load while maintaining feature matching accuracy.
Solution Approach 2:
The patent applies different data inclusion criteria to different regions of the target article. Measurable surfaces are included in the model data with full detail, while recessed areas or regions that cannot be captured by the measuring machine are excluded or represented differently, creating a model with non-uniform data quality that matches the actual measurement capabilities.
2Reliability
If feature points from both model data and actual measurement data are matched, then position posture identification becomes impossible when measurement data is incomplete
Solution Approach 1:
The patent creates a simplified model data copy that replicates only the measurable portions of the target article. This copied model excludes features that cannot be captured by the measuring machine, allowing feature point matching to proceed reliably by comparing actual measurement data against a model that accurately reflects what can be measured.
Solution Approach 2:
The patent uses a partial model that includes only the necessary measurable features rather than attempting to represent the complete object. This partial action approach focuses computational resources on matching the features that actually exist in the measurement data, avoiding the impossibility of matching features that were never captured.
3Ease of manufacture
If a single three-dimensional measuring machine is used, then system cost is reduced, but measurement coverage becomes insufficient
Solution Approach 1:
The patent segments the target article's surface into measurable and non-measurable regions based on the single measuring machine's capabilities. By dividing the model data into regions that can be captured (visible surfaces) and regions that cannot (recessed areas), the system makes effective use of limited measurement coverage while maintaining accurate position posture identification for the measurable portions.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach efficiently identifies the position posture of a target article by reducing processing loads and improving accuracy by omitting or weighting data that cannot be measured, thereby enhancing the reliability of posture identification in robot systems.
Implementation Method 1
the three-dimensional measuring machine shoots, with a camera, a pattern obtained by projecting pattern light from a projector, and point cloud data within a measurement region is acquired from the position of the pattern on an image acquired with the camera by the principles of a stereo measurement
Data Source
AI summary
A position posture identification device includes: an actual measurement data acquisition unit which acquires actual measurement data on the shape of a workpiece; a virtual measurement data generation unit which generates, from shape data defined for the workpiece, virtual measurement data; a filter processing unit which performs, based on the measurement property of the three-dimensional measuring machine, affine transformation on the virtual measurement data; a feature point extraction unit which extracts feature point data from the actual measurement data and the virtual measurement data; a storage unit which stores, as model data of the workpiece, the feature point data extracted from the virtual measurement data; and a position posture calculation unit which checks the feature point data of the actual measurement data against data obtained by performing coordinate transformation on the feature point data included in the model data so as to calculate the position posture of the workpiece.