Machine Learning | Mainblades Aircraft Drone Inspections

Why does Machine Learning matter?

In the last few years, Artificial Intelligence (AI) and its subset Machine Learning (ML) have gained ground in numerous industries. Aviation is no exception.

Mainblades uses it to disrupt the way aircraft inspections are conducted, turning a tedious and hazardous task into an automated and efficient process.

It’s an exciting time to be in this industry because there is scope for more – more innovations, more game-changing results!

Aircraft inspections through AI

It is no secret that the use of drone technology is the latest trend gaining traction in the aviation industry. Using drones to inspect aircraft is becoming sought-after among lessors, airlines, and Maintenance, Repair & Overhaul (MRO) organizations, thanks to the latest work carried out by companies such as Mainblades.

We combine state-of-the-art drones, smart algorithms & ML to provide an out-of-the-box solution for aircraft inspections. Our drone’s ability to identify damages in complex real-world environments makes it the right tool for visual aircraft inspections.The automated processing and analysis of data is a crucial element. This is possible by what we call a Machine Learning Stack, with data acquisition as its most important component.

How to turn data into actionable insights?

Data Acquistion

The quality of an ML algorithm depends on the quality of the data you feed to it. That is why the data we collect and use to develop our models is diverse. We collect aircraft damages that can occur during service, such as
- dents
- buckles
- lightning strikes
- bird strikes
We cover all possible situations including different aircraft types, locations as well as various environmental and lighting conditions during the inspection.
Additionally, we use the domain knowledge of our customers in tight collaboration with machine intelligence. Thus, we rely on aircraft engineers’ experience and expertise to improve our model for detecting deviations and adapt it to their needs.

damage recognition through machine learning

Model development

While the drone is flying around the aircraft taking high-resolution images, our machine learning model makes predictions in the iPad application (boxes drawn around the damages).
Our team uses these insights and data inputs to further optimize the algorithm. This way, the model learns over time to perform object and damage detection, predicting its location and type only with the help of images.
By exploiting the recent advancements in AI and ML, today the drone is able to fly indoors and outdoors, inspect every aircraft type and detect damages of about 2mm size.

group of people during aircraft inspection

Using the insights

The ML-based damage detection is translated into a representation that is meaningful and intuitive to the inspector and adheres to industry standards (e.g., Aircraft Maintenance Manuals).
The inspector gets the chance to make a final assessement. He can either accept or reject these predictions, change their shape or add new ones.
As a result, reports are created in a much faster, easier and safer manner. Data can be shared in real-time because the data is immediately uploaded to our cloud-based system. This way, aircraft engineers can make informed decisions in no time.