Publication:

Digital twin development for VTOL UAVs

The use of vertical take-off and landing (VTOL) drones in various industries is becoming increasingly popular. However, the democratization of drones also raíces significant concerns about safety during their operations. Their relatively new technology and growing interest often lead to the use of undertested vehicles in missions where safety is critical. To address these risks and reduce the costs associated with experimental flights of new aircraft, the implementation of high-fidelity emulators is proposed. These advanced emulators, known as digital twins, have seen exponential growth in popularity in recent years. This thesis addresses the complete process of developing digital twins for VTOL unmanned aerial vehicles (UAVs). Initially, we propose a digital twin development model based on a variation of the double diamond design process. After identifying critical systems in two types of VTOL UAVs, a hexacopter and a commercial tiltrotor aircraft, we have developed a mathematical model to characterize aerodynamic, gravitational, and propulsive actions. Propulsive actions are measured through experimental tests on a motor test bench. Gravitational actions are determined using precise computeraided design/computer-aided manufacturing (CAD/CAM) models and experi mental measurements. Finally, aerodynamic actions are obtained through a novel aerodynamic model, which calculates the complete aircraft aerodynamics based on the incident wind direction, relying on numerous computational fluid dynamics (CFD) simulations. To mitigate the costs associated with CFD simulations, we employ surrogate models, developing and validating a surrogate model capable of potentially reducing the number of simulations by up to 50%. Another critical subsystem is the communication system. Based on experimental measurements, we fit and validate a path loss model for Received Signal Strength Indicator (RSSI) to estimate signal losses as a function of the aircraft’s position and attitude. This model allows us to predict and quantify the impact of the UAV’s attitude and position relative to the communication source. Ultimately, the digital twin is successfully implemented and validated using both hardware-in-the-loop and X-Plane for a commercial flight controller, as well as software-in-the-loop and Gazebo for an open-source controller (Ardupilot). This comprehensive validation approach ensures that the digital twin accurately replicates the behavior and performance of the actual UAV systems, regardless of the emulation engine and architecture chosen.