Loss of Control (LOC), a common form of failure in unmanned aerial systems, has recently gained considerable attention. Indeed, with the projected increase in drone usage, safety has become a critical element. Whereas most of the current research focuses on addressing a particular challenge (e.g., partial actuator failure), a global solution is required when a vehicle undergoes an unknown amount of damage. This thesis work looks into one technique called “Critical Slowing Down.” This data-based technique is commonly used in biological systems to detect early warnings before a critical transition by measuring changes in statistical indicators. The method was tested on both an inverted pendulum on cart simulation and real damaged quadcopter flight data. Reliable Early Warning Signals (EWS) were found in both cases – especially when the actuator saturation frequency is considered. However, more research needs to be done before implementing this method on a real system. The robustness of these EWS must be improved, and a relationship between these precursors and the time left before LOC must be derived.