Design and implementation of SMO for a nonlinear MIMO AHRS

In a low-cost attitude heading reference system (AHRS), the measurements made by MEMS inertial and magnetic sensors are affected by large parameter uncertainties, stochastic noises and unknown disturbances. In this paper, considering the robustness of the sliding mode observers (SMO) against both structured and unstructured uncertainties as well as exogenous inputs, the process of design and implementation of a nonlinear SMO is proposed for a low-cost AHRS. For simultaneous estimation of orientation variables and calibration biases of gyroscopes, a nonlinear and non-affine model of the AHRS is considered. Therefore, based on the Lie-algebraic method, the estimation algorithm is designed for a general class of non-affine nonlinear MIMO systems. In the proposed observer, owing to decreasing the required assumptions for coordinate transformation in recent literatures, the design process of the SMO is simplified. The gain matrices of the proposed SMO are obtained through ensuring the stability and the convergence of estimation errors based on Lyapunov's direct method. The expected tracking performance of the robust state and parameter estimation algorithm compared to that of the extended Kalman filter (EKF) is evaluated through simulations and real experiments of a strapped AHRS on a ground vehicle.

► Based on Lie-algebraic, a SMO is designed for nonlinear and non-affine MIMO systems.
► Stability and convergence of estimation errors are guaranteed by Lyapunov's method.
► The proposed sliding mode observer is implemented in a MEMS AHRS.
► Simulation and real test results show that our SMO is superior to the EKF.