This thesis explores a unique convergence of aerial robotics, wireless communication, and advancedmachine learning regression algorithms to develop a process pipeline method for developing an autonomous localization algorithm using only a spectrum-efficient narrow-band transmission. In the face of increasing demands for narrow-band wireless spectrum usage, this work gives a conjunctive method that does not rely on traditional wireless localization techniques such as wideband signals, multiple antennas, or precise timing. Instead, this approach capitalizes on the stationarity of the target and its unique pilot radio signal. At the same time, this approach emphasizes the rapid localization of the signal transmission within ten minutes and across a 20-acre area. A Gaussian process and Bayesian optimization are used to find the transmission signal’s path-loss function non-parametrically and map it to a 2D spatial field from samples collected as the drone completes its task. This research becomes foundational in settings where Unmanned Aerial Vehicles (UAVs) are becoming increasingly vital for applications in search and rescue, last-mile delivery, and warfare, necessitating sophisticated, capable, and efficient algorithms.