Detecting drone manufacturing defects by an AI real-time early warning system

The overall goal of the experiment is to develop an AI real-time early warning system to detect defects during the last stage of our drone manufacturing process: the validation flights. The objectives of this system are:

• Reduce the number of aircraft crashes caused by major defects by warning flight crews that the flight parameters are not correct before the defects can generate an accident.
• Reduce the undetected defects that can generate problems (need for maintenance/accidents) after the aircraft is delivered to clients and is performing real operations.
• Generate a flight data space prototype where the data of the validation flights are stored and analysed using AI to continuously improve the system’s detection capability of defects. This data space will be further developed to include all the flight logs

As a result of achieving these objectives, a derived goal of the project is to provide EASA (European Aviation Safety Agency) with additional data that demonstrate the safety of the drones we manufacture.

This project aims to develop an AI cloud-based system for Manufacturing Applications enabling human-AI collaboration to promote a zero-defect policy in a drone manufacturing SME. This experiment is aligned with the goals of this project (Human-AI interaction, collaborative intelligence). This project is also aligned with Basque Country’s DIH manufacturing domain (AI-enabled Platform for Zero Defect). Moreover, although we are applying to Topic-2, the experiment will generate our flight validation data space embryo, where all the flight data will be stored.
The project will contribute to the AI4EU with datasets and an use-case of the use of AI solutions to guarantee and demonstrate safety in autonomous vehicle manufacturing and development.