Unmanned aerial vehicles have become increasingly used in recent decades, and the applications that use this type of robot are diverse. Knowing the state of the battery of these vehicles is also of great importance, along with the possibility of estimating it. There are many methods for approximating the voltage of a battery, from model-based to data-driven models. This paper presents the creation, analysis, and testing of several data-based models for the battery of a Crazyflie 2.0 drone. The goal is to estimate battery voltage based on planned trajectories using machine learning algorithms. The work implements a variety of machine learning models, such as Decision Trees, Random Forests, and Support Vector Machines, created using MATLAB. A sequence of flights yields experimental results, with data acquisition focusing on the drone’s position and the battery’s voltage. The machine learning algorithms are compared based on the prediction error.