ANN-Based Tilt Angle Prediction for Improved UAV Autonomous Flight

This study aims to develop an Artificial Neural Network (ANN) based model to predict the tilt angle of Unmanned Aerial Vehicles (UAVs) by considering external forces such as wind speed. With the increasing commercialization of self-driving UAVs for various applications, including shipping, cleaning, and delivery, ensuring their safe and efficient operation is crucial. GPS signals are not always reliable during autonomous flight, which can lead to collisions or accidents. The proposed ANN model aims to reduce the cost and complexity of UAV control systems while maintaining high prediction accuracy (
$R^{2}$
values of 0.8592, 0.8722, and 0.9678). By using data sets obtained from current and past states, the model can predict future states and stably control the UAV without deviating from its path. Although this study focuses on wind speed as the primary external force, incorporating additional data sources such as gyro sensors, temperature, barometric pressure, and image data could further enhance the accuracy of tilt angle predictions and self-flying control. The proposed model has the potential to be applied to various UAVs and provide critical information for decision-making in autonomous flight.