A Generalized Reduced-Complexity Inertial Navigation System for Unmanned Aerial Vehicles

In this paper, a generic approach to attitude and position estimation, suited for any type of unmanned aerial vehicle, is presented. This will be achieved by establishing a generic framework, which can be extended using adaptive methods to determine the thrust properties of the engines and the mass of the aircraft, while keeping the overall computational complexity of the system low. Furthermore, the effect of magnetic disturbances will be reduced in a novel way by confining the magnetic errors to affect only heading, without compromising the pitch and roll estimation of the system with error-based estimation. The efficacy of the proposed framework will be evaluated through extended simulations and experimental validations on a multirotor. Finally, guidelines will be provided toward: 1) an implementation with a reduced computational complexity and 2) the utilization of the square-root formulations of the extended Kalman filter for extending the dynamic range of the filter.