In recent years, deployment of drones for logistics and passenger transportation has garnered significant attention. Accurate estimation of battery remaining capacity and flight duration is essential for safe operation. This study proposes a method to predict drone power consumption using a surrogate model specifically, Gaussian process regression without relying on detailed physical models. Simulation data generated under various flight conditions and environmental scenarios were used to train the model. The results of theoretical evaluation demonstrated that the model could predict power consumption within approximately 3% error margin. This approach enables real-time estimation of remaining flight time under diverse conditions, contributing to enhanced safety and reliability in drone operations.
Okuda et al. (Wed,) studied this question.