Object Location on Pallets via Edge Distance Frequency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for identifying the location of rectangular objects on pallets in automated warehousing are inefficient, particularly when the size of the objects is unknown, as they struggle to distinguish between noise and object edges using line-finding or corner-finding algorithms.
Innovation Solution
A system that analyzes a 2D image of a pallet using gradient evaluation to detect edge points, determine the dimensions and center coordinates of 3D rectangular objects by counting frequency distances between edge points, and identify positions with a concentration of edge points at half the object's size, enabling accurate object location and removal.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If line-finding or corner-finding algorithms are used to identify object locations, then automated warehousing operations can be performed, but the system cannot reliably distinguish between noise and object edges when object sizes are unknown
Solution Approach 1:
The patent changes the detection parameter from absolute position (line-finding) or corner coordinates (corner-finding) to relative distance measurements between detected edges. By measuring distances between adjacent edges and analyzing frequency distributions, the system can identify object boundaries regardless of noise, since true object edges will consistently produce distance measurements matching known object dimensions while noise will not.
2Adaptability or versatility
If the system attempts to identify objects of varying sizes and orientations, then versatility is improved, but the complexity of distinguishing objects from noise increases
Solution Approach 1:
The patent creates a frequency distribution histogram that copies and aggregates distance measurement data from multiple edge detections. By building this statistical representation of expected object dimensions, the system can compare new detections against the established pattern, automatically adapting to various object sizes and orientations without requiring complex predefined models for each object type.
Data Source
AI summary
Methods are provided for locating 3D rectangular objects on surfaces, each object having substantially the same X, Y and Z dimensions as the other objects on the surface. An image of a surface having objects arranged thereon is obtained. The image includes pixels arranged in rows and columns. As each object has the same Z-dimension, the surface shown in the image is planar. From the image, the X and Y dimensions, and X, Y coordinates of a center for each object are determined. Edge points are detected utilizing gradient evaluation. The X and Y dimensions of the objects are determined by counting the frequency of distances between edge points and X, Y coordinates of the center location for each object is determined by identifying positions in the image that have a concentration of edge points at a distance equivalent to one-half the size (X or Y dimension) of the objects.