Geometric Feature Grouping for 3D Pose Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for measuring the position and orientation of objects using geometric features in robotics are hindered by increased processing costs due to grouping processing, which impairs speed and efficiency in model fitting.
Innovation Solution
A technique that involves generating viewpoint-dependent position and orientation estimation models by grouping geometric features based on their influence on six degrees of freedom parameters, allowing for efficient selection and use of features with diverse influences, thereby reducing processing costs and improving estimation speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If grouping processing of geometric features is executed during model fitting processing, then geometric features can be organized based on their influence on position and orientation parameters, but processing cost increases and speed of model fitting processing is impaired
Solution Approach 1:
The patent performs grouping processing of geometric features in advance during model generation, rather than during model fitting processing. The grouping unit organizes geometric features into groups based on their influence on position and orientation parameters before the actual measurement task. This preliminary organization allows the measurement apparatus to directly utilize pre-grouped features during model fitting, eliminating the computational overhead of grouping operations during the time-critical measurement phase and thereby maintaining high processing speed while ensuring accurate feature selection.
2Measurement precision
If all geometric features are used in model fitting, then comprehensive information is captured, but processing time increases due to the large number of features
Solution Approach 1:
The patent segments geometric features into distinct groups based on their influence characteristics on position and orientation parameters. The grouping unit divides the complete set of geometric features into multiple groups, where each group contains features with similar influence patterns. During model fitting, the system can then select and process only the necessary groups or representative features from each group, rather than processing all individual features. This segmentation approach reduces the effective number of features processed while preserving the comprehensive information needed for accurate position and orientation estimation.
Solution Approach 2:
The patent applies partial action by selecting only the necessary geometric feature groups for model fitting based on their influence on position and orientation parameters. Instead of processing all geometric features, the system identifies and processes a subset of features that provide sufficient information for accurate measurement. The grouping structure enables the system to determine which feature groups are most relevant for the current measurement task, allowing selective processing that reduces computational time while maintaining measurement accuracy through the use of strategically selected representative features.
Data Source
AI summary
A three-dimensional shape model of a target object is input. A position and orientation of at least one image sensing device used to capture an image of the target object is set so as to virtually set a relative position and orientation between the target object and the image sensing device for the three-dimensional shape model of the target object. At least one position and orientation of the image sensing device is selected. Geometric features are grouped based on a relationship between an image to be obtained at the selected position and orientation of the image sensing device and the geometric features of the three-dimensional shape model.