Robot Position Detection Using Precomputed Feature Value Tables

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing object detecting apparatuses using three-dimensional shape models require significant time to accurately calculate the position and orientation of objects due to the use of six variables in a three-dimensional Cartesian coordinate system.

Innovation Solution

A robot apparatus that generates and stores a feature value table associating position and orientation data with reference model images, allowing for the calculation of object position and orientation using feature values from photographed images, thereby reducing the need for heavy-load calculation processing.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If six variables indicating translation and rotation in a three-dimensional Cartesian coordinate system are used to calculate the position and orientation of the target object, then the accuracy of position and orientation detection is improved, but the detection time increases enormously

Engineering Contradiction:
Improveposition and orientation detection accuracyVSAvoiddetection time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent pre-calculates and stores feature values for multiple predetermined positions and orientations of the reference model in a feature value table before actual detection. During detection, the system only needs to compare the captured image's feature value against the pre-stored values, eliminating the need for real-time six-variable calculations and dramatically reducing detection time while maintaining accuracy.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent creates a feature value table that contains pre-computed feature values representing the relationship between image features and six-variable parameters for various positions and orientations. This copied feature value table serves as a lookup reference, allowing the system to determine position and orientation by simple comparison rather than complex real-time calculation.

Inventive Principle:
Principle #26Copying

2Measurement precision

If a three-dimensional shape model is applied to estimate the position and orientation of the target object, then the detection accuracy is improved, but the calculation processing becomes heavy-load and time-consuming

Engineering Contradiction:
Improveposition and orientation estimation accuracyVSAvoidcalculation processing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system performs the complex three-dimensional model calculations in advance, generating a feature value table that stores the relationship between image feature values and six-variable parameters for multiple predetermined positions and orientations. This pre-processing eliminates the need for heavy real-time calculations during actual detection, reducing computational complexity while maintaining the accuracy benefits of three-dimensional modeling.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent replaces the mechanical computation process of real-time six-variable calculation with a data lookup process using the pre-generated feature value table. Instead of performing complex mathematical operations during detection, the system substitutes this with simpler comparison and retrieval operations, significantly reducing computational complexity.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS9082017B2Robot apparatus and position and orientation detecting method
Publication Date: 2015.07.14 SEIKO EPSON CORP
  • US9082017B2 patent drawing
  • US9082017B2 patent drawing
  • US9082017B2 patent drawing

AI summary

A robot apparatus includes a reference-model storing unit configured to store a reference model of an object, a feature-value-table storing unit configured to store a feature value table that associates position data and orientation data of the reference model and a feature value, a photographed-image acquiring unit configured to capture a photographed image of the object, a detecting unit configured to calculate a photographed image feature value from the photographed image, and a driving control unit configured to control a robot main body on the basis of the position data and the orientation data to change the position and the orientation of a gripping unit.