Enhancing Data Discretization for Smoother Drone Input Using GAN-Based IMU Data Augmentation

Abstract

This study investigates the use of generative adversarial network (GAN)-based data augmentation to enhance data discretization for smoother drone input. The goal is to improve unmanned aerial vehicles’ (UAVs) performance and maneuverability by incorporating synthetic inertial measurement unit (IMU) data. The GAN model is employed to generate synthetic IMU data that closely resemble real-world IMU measurements. The methodology involves training the GAN model using a dataset of real IMU data and then using the trained model to generate synthetic IMU data. The generated synthetic data are then combined with the real data for data discretization. The resulting improved data discretization is evaluated using statistical metrics and a similarity evaluation. The improved data discretization demonstrates enhanced drone performance in terms of flight stability, control accuracy, and smoothness of movements when compared to standard data discretization methods. These results highlight the potential of GAN-based data augmentation for enhancing data discretization and improving drone performance. The proposition of improved data discretization offers a tangible benefit for the successful integration of Advanced Air Mobility (AAM) systems. Enhancing the accuracy and reliability of data acquisition and processing in UAS makes UAS operations safer and more reliable. This improvement is crucial for achieving the goal of automated and autonomous operations in diverse settlement environments, encompassing multiple mobility modes such as ground and air transportation.