Multistage-Fusion Algorithm for Estimation of Aerodynamic Angles in Mini Aerial Vehicle

Abstract

Nature is the best teacher of many scientific research and development. One such nature inspired invention is the flight vehicle system. Among these, Mini Aerial Vehicles (MAVs) are a class of flight vehicles which have size close to birds. The current work is inspired by the fact that MAV with Low Aspect Ratio (LAR) wings (Aspect Ratio < 2) has nonlinear lift curves and it does not stall sharply as compared to high aspect ratio wings. This can be utilized in flying the MAVs at lower speeds and at higher angle of attacks. In LAR wings, the lift is constant or keeps increasing up to large Angle of Attack (AOA) which can help in flying the MAV at slow speed. Main focus of the present work is to estimate the AOA, which can be further used for development of control laws. This paper presents a data fusion algorithm for estimating the aerodynamic angles in a MAV. For simulation, true states of the aircraft motion are generated using a flight simulation program and a zero mean white noise is added to few of these states as required for the measurements. These noisy states are used as sensor measurements for estimating the AOA and Side Slip Angle (SSA). The proposed scheme is a multi-stage in which initially Euler angles are estimated and later stages are used for AOA and SSA estimation. In the first stage, the Euler angles are estimated using the accelerometer outputs and the rate gyro outputs and V in an Extended Kalman Filter (EKF) algorithm. As a first attempt for estimation of AOA and SSA, only acceleration, angular rates and airspeed were used as measurement in six states EKF. It was observed that there is a bias of more than 10° ° ° ° present in the estimated aerodynamic angles. Since, it is not possible to have any additional sensors on-board due to weight restriction, a new modification is proposed in which a pseudo estimation of AOA and SSA was used as measurement. This effectively reduces the estimation bias and the mean error in both AOA and SSA, and are below 1°. Nomenclature Fx = Total force along the body x-axis Fy = Total force along the body y-axis Fz = Total force along the body z-axis L = Total moment along the body x-axis M = Total moment along the body y-axis N = Total moment along the body z-axis p = Roll rate q = Pitch rate r = Yaw rate Cx = Force coefficient in the x direction Cy = Force coefficient in the y direction Cz = Force coefficient in the z direction Cl = Moment coefficient in the x direction Cm = Moment coefficient in the y direction