Data-driven State Awareness for Fly-by-feel Aerial Vehicles: Experimental Assessment of a Non-parametric Probabilistic Stall Detection Approach

Abstract

In this work, an experimental study of a novel data-driven fly-by-feel state awareness method is presented. A non-parametric probabilistic approach for stall detection is investigated and assessed via a series of wind tunnel experiments. The method is based on the statistical analysis of the recorded signals and subsequent statistical hypothesis testing and decision making procedures. In this proof-of-concept experimental study, the flight state is defined by two variables, the airspeed and angle of attack. The experimental evaluation and assessment is based on a prototype bio-inspired self-sensing composite wing that is subjected to a series of wind tunnel experiments under multiple flight states. Distributed micro-sensors, in the form of stretchable sensor networks, are embedded in the composite layup of the wing in order to provide the sensing capabilities. Experimental data collected from piezoelectric sensors are employed for the development and assessment of a non-parametric fly-by-feel stall detection approach within a probabilistic framework. In this study, special emphasis is given to the early detection of aerodynamic stall without making use of any conventional information related to the attitude of the vehicle. The method is able to provide in real time the probability of stall for an indicative flight scenario that was implemented in the wind tunnel. The obtained results demonstrate the effectiveness and potential of the developed approach