Description

Within the framework of this task, the main examples of the use of recognition and classification technologies in the industrial process using neural networks and fuzzy logic are considered. This includes: classification of an object according to its composition, search for the optimal composition, the task of controlling the stand’s manipulators, etc.
This example considers working with objects on a conveyor belt: recognizing and positioning an object on the belt using cameras, finding object defects, and controlling a mechatronic manipulator (industrial robot) based on information received from optical sensors.
This hardware and software complex consists of software designed to obtain synthetically generated images of products on a conveyor belt in a virtual space completely similar to the real one. The images obtained in the virtual studio can be used to form samples and apply machine learning methods to classify objects, as well as to control the quality of synthetic samples; the data obtained can also be used to solve the problem of controlling a mechatronic manipulator in order to move objects in space (removing defects from tapes).
• The physical part of the complex consists of a model small-sized conveyor, as well as a set of specialized cameras for fixing samples and collecting data.
• This complex is designed to test the basic mechanisms for processing signals from optical sensors to form a sample for machine learning of neural networks and subsequent image classification.