3D Pose Recognition via 2D Template Matching for Thin-Edge Containers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Industrial robots face challenges in recognizing the three-dimensional position and posture of containers with thin edges due to difficulties in obtaining three-dimensional point group information, leading to potential collisions and damage.
Innovation Solution
An information processing device generates a two-dimensional template from three-dimensional model information, allowing for accurate recognition of the object's position and posture by matching a captured image with the template, even for objects with small widths or specular light-reflective materials.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If three-dimensional point group information is acquired using a distance sensor, then the position and posture of box-shaped workpieces can be recognized, but it becomes difficult to acquire accurate information for objects with thin edges or small widths due to light reflection issues
Solution Approach 1:
The patent replaces the distance sensor-based three-dimensional measurement system with a two-dimensional image processing system. Instead of using optical ranging that fails on thin edges, the invention uses a camera to capture two-dimensional images and processes them through template matching and coordinate transformation to achieve accurate position and posture recognition, thereby substituting a measurement method that works reliably for all object types.
Solution Approach 2:
The patent transforms the three-dimensional measurement problem into a two-dimensional image processing problem. By capturing the object in two-dimensional space and using template matching algorithms, the system can accurately determine position and posture without relying on three-dimensional point cloud data that is difficult to acquire for thin-edged objects. The coordinate relationship unit then maps these two-dimensional results back to three-dimensional coordinates.
2Measurement precision
If traditional three-dimensional measurement methods are used, then position recognition can be achieved, but computational complexity increases and accuracy decreases for objects with specular light-reflective materials or small widths
Solution Approach 1:
The patent extracts only the essential two-dimensional projection information from three-dimensional objects, ignoring the complexity of full three-dimensional surface mapping. By focusing on two-dimensional template matching rather than complete three-dimensional point cloud processing, the system reduces computational complexity while maintaining accuracy for position and posture recognition.
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
Provided is an information processing device that can recognize, with a high degree of accuracy, a three-dimensional position/posture. The present invention comprises: a first input unit that receives the input of three-dimensional model information for generating a three-dimensional model of an object to be recognized; a template generation unit that generates, on the basis of the three-dimensional model information, a two-dimensional template which shows the shape of the upper edge of the object to be recognized; a coordinate relationship identification unit that generates coordinate relationship information indicating the relationship between three-dimensional coordinates of the three-dimensional model and two-dimensional coordinates when the three-dimensional model is imaged; a second input unit that receives the input of the captured image in which the object to be recognized has been captured; a matching unit that matches the captured image and the two-dimensional template; and a recognition unit that recognizes the three-dimensional position and posture of the object to be recognized by referencing the coordinate relationship information for the object to be recognized in the captured image, which is detected according to the results of the two-dimensional matching by the matching unit.