This paper presents an automated workflow for antenna radiation pattern characterization based on an autonomous Unmanned Aerial Vehicle (UAV) or drone. To this end, first a radio frequency payload is fitted in a gimbal and mounted on the UAV to simulate signals received by an Antenna Under Test (AUT) from satellites. Afterward, a spectrum analyzer is connected to the antenna system and records the strength of the signals. The data are then used to find antenna's main beam location and radiation pattern characterization based on Bayesian optimization for real-time path planning. Our proposed approach accelerates localization of important features in antenna testing, such as radiation pattern can be identified without the need of post-processing. Furthermore, the process of scanning can be done in shorter distance, thereby, saving drone batteries.*This work was supported by QuadSAT, a company that supplies dronebased antenna testing and tracking solutions in Denmark and partially funded by Equinor's gift professorship at NTNU.