Opto-Electro-Robotics Lab.


Artificial Intelligence base visual inspection
Our AI research team has been developing an automatic system that recognizes and classifies a damage in the complicated structures, such as an aircraft engine. In detail, there are two research topics:

* A vision-based system for a damage detection
* Visual Inspection by drones and AI

* Scanning free space measurement system (SFSM system)

Machine learning teaches computers to do what comes naturally to humans and animals: learn from experience. Machine learning algorithms use computational methods to “learn” information directly from data without relying on a predetermined equation as a model. The algorithms adaptively improve their performance as the number of samples available for learning increases.

In our cases, machine learning is used for damage classification and recognition. The machine learns from images of intact and damaged structures. Therefore, the most appropriate algorithm is a convolutional neural network (CNN) because it takes an image as an input directly without any modification. Until now, we have only dealt with image data, but it could be expanded to any other data types. :

* Schematic diagram of CNN

* Engine fan blade damage matching /Engine fan blade damage recognition