Globally asymptotic attitude estimation for accelerated aerial vehicles

This paper addresses the attitude estimation problem for rigid bodies moving in 3D-space with not negligible and time varying acceleration. For this case, we propose a simple and efficient globally asymptotic attitude estimation by using measurements from low-cost IMU and GPS. It consists of an observer based on proportional-integral (PI) term with discontinuous switching function to estimate the linear acceleration in the inertial frame for which we obtain global exponential convergence, along with a complementary filter for raw vectors from IMU to remove noise and bias. For this one, we get global asymptotic convergence of the observer taking into account gyro bias. Thereafter the filtered data is processed with an algebraic algorithm (e.g. TRIAD) for the attitude estimation. Note that when no gyro bias is considered, we obtain global exponential convergence for the overall system. To show the importance of filtering these raw vectors and the convergence behaviour of the proposed observers, simulation results are also given.