Recently, a team of researchers from South Korea aimed at a multi-rotor vehicle that can self-assess its performance and instantly identify and resolve propulsion faults by developing a data-driven diagnostic approach centered on the IMU. The methodology marks a paradigm shift in UAV maintenance by leapfrogging from mere routine visual inspections to sophisticated diagnostic nuances.
Unlike traditional fault diagnosis methods that rely on additional sensors and RPM measurements, the investigated method utilizes Principal Component Analysis (PCA) of IMU signals (accelerometers and gyroscopes) to interpret fault data. This method excels at handling noisy data without the need for additional sensors and therefore without additional costs. The principal direction vectors derived from the IMU data are then applied to a supervised learning algorithm that not only detects faults but also measures their severity and location. It is this strategic application of IMU data processed through novel algorithms that makes IMUs an indispensable tool in UAV diagnostic techniques.
The essence of this research is not only to diagnose but also to predict the actuator health of a multi-rotor vehicle. This approach marks a shift in multi-rotor vehicle operations, providing engineers with a smarter and more resilient framework for UAVs.