Robot Vision Learning for Variable Workpiece Packing Ratios
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing robot systems face challenges in stably obtaining information about workpieces that are randomly piled up, as the appearance of workpieces varies significantly with different packing ratios, making it difficult to accurately detect their positions and orientations.
Innovation Solution
The method involves generating a learned model using machine learning with diverse data sets that include images of workpieces at various packing ratios, allowing the robot system to adapt to different situations and accurately detect workpiece information regardless of the packing density.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If machine learning is performed using image data with a fixed number of workpieces, then the learning model can be trained efficiently, but the detection accuracy deteriorates when the actual number of workpieces varies
Solution Approach 1:
The patent applies parameter changes by varying the number of workpieces in the training image data. Multiple sets of image data are prepared with different numbers of workpieces (first number, second number, etc.), and the machine learning model is trained on this diverse data. This enables the model to adapt to varying workpiece quantities and maintain high detection accuracy regardless of the actual number of workpieces in the container.
2Device complexity
If the robot system uses a simple detection method, then the system complexity is reduced, but the ability to handle varying packing ratios deteriorates
Solution Approach 1:
The patent uses virtual workpieces and virtual container data as copies to generate diverse training data without requiring physical manipulation of actual workpieces. Virtual image data with varying numbers and arrangements of workpieces is created through simulation, enabling the machine learning model to learn different packing scenarios efficiently while keeping the physical system simple.
Data Source
AI summary
An information processing method for obtaining a learned model configured to output information of a workpiece includes obtaining first image data and second image data. The first image data includes an image corresponding to a first number of workpieces disposed in a container or to the first number of virtual workpieces disposed in a virtual container. The second image data includes an image corresponding to a second number of workpieces disposed in the container or to the second number of virtual workpieces disposed in the virtual container. The second number is different from the first number. The information processing method includes obtaining the learned model by machine learning using the first image data and the second image data as input data.