Robust UAV Attitude Estimation Using a Cascade of Nonlinear Observer and Linearized Kalman Filter

This paper presents a new approach for Unmanned Aerial Vehicle (UAV) attitude estimation using a cascade of nonlinear observer and linearized Kalman filter. The nonlinear observer is globally asymptotically stable and is designed using linear matrix inequalities (LMI). The exogenous signal from the nonlinear observer is used to generate a linearized model for the Kalman filter. The method is implemented for attitude estimation of a quadcopter. The nonlinear model is derived from the Newton-Euler equations. The nonlinear model is locally Lipschitz due to the cross and dot products between the angular and linear velocity vectors. The attitude estimation from the dynamical system presented in this paper can be used as a module for fault detection. Simulations in Gazebo on a PX4 using Software In The Loop (SITL) shows the proposed method is able to estimate the attitude of a quadcopter accurately.