Robot Article Extraction Using 3D Point Projection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing article take-out apparatuses face difficulties in differentiating between the open end face or inside walls of a container and the articles stored inside, leading to mistaken recognition of 3D points.
Innovation Solution
An article take-out apparatus that uses a 3D measuring device to measure surface positions of articles within a container, a camera to image the open end face, and a control device to set a search region on the image, project 3D points onto a reference plane, and judge whether these points are within the search region, preventing mistaken recognition of container walls as articles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If a camera is used to take an image of the entire container including open end face and inside walls, then the field of view is sufficient to detect all articles, but differentiation between container structures and articles becomes difficult
Solution Approach 1:
The patent divides the image processing into two distinct stages: first capturing a wide-field image to identify the container's open end face and establish a search region, then using a zoomed-in image of only the storage space to accurately detect articles. This segmentation separates the container structure detection from the article detection, resolving the contradiction between wide coverage and precise recognition.
Solution Approach 2:
The patent transitions from a single 2D image to a multi-dimensional approach by capturing images at different zoom levels and processing them in sequence. The first image provides spatial context (open end face location), while the second zoomed image provides detailed article information, effectively adding a dimensional aspect of magnification to the detection process.
2Quantity of substance
If image processing is performed on the entire container image, then all potential article locations are covered, but computation time increases and accuracy decreases due to container structure interference
Solution Approach 1:
The patent extracts and isolates the storage space region from the entire container image by first identifying the open end face and calculating the search region boundaries. This extraction removes irrelevant container structure data (inside walls, open end face) from the processing scope, reducing computation time and preventing misidentification while maintaining complete article detection coverage.
Solution Approach 2:
The patent performs preliminary actions by first capturing and processing the wide-field image to identify the container's open end face and establish the search region before performing article detection. This preliminary region definition prepares the system by pre-defining where articles are likely to be found, so the subsequent detailed detection can focus only on relevant areas, reducing overall processing time.
Data Source
AI summary
An article take-out apparatus including, acquiring a reference container image including an open end face of a container by imaging operation by an camera, setting an image search region corresponding to a storage space of the container based on the reference container image, setting a reference plane including the open end face of the container, calculating a search region corresponding to the image search region based on a calibration data of the camera stored in advance, converting the search region to a converted search region, taking out 3D points included in the converted search region by projecting a plurality of 3D points measured by the 3D measuring device on the reference plane, and recognizing positions of articles inside the container using the 3D points.