Flying Robots with Wind Awareness: Estimation of the Wind Velocity and Direction through the Data Log and a Closed-Form Energy Model

Abstract

Flying robots must predict their remaining flight time at each time a decision should be taken. Estimating aircraft states is a challenging problem, especially if the aircraft is a light and small Unmanned Aerial Vehicle (UAV), under the effect of strong winds. In those cases, the energy consumption can vary in such a way that it would directly compromise the mission. This paper proposes a method for estimating the wind velocity and direction based on how much energy the aircraft needs to fly a predetermined path. The energy is calculated using a closed-form energy model based on the dynamic of the movement, the principles of superposition, and energy conservation. The model considers each parcel of energy consumed separately in the equation, even the drag force that is normally despised in other models. This model should be extended for other types of multi-rotors because it is a function of quadcopter parameters. The model has been validated against another model published recently in the literature. Here the model is applied in a reverse form. The consumed energy estimated by a Kalman Filter is applied as an input to the model such that the velocity of the quadcopter relative to the mass of air is calculated. That data and information supplied only by the flight computer allow determining the wind parameters. Despite the noisy characteristics of wind, it works properly, and the results demonstrate the feasibility of the proposed approach.