Design and Real Time Implementation of a Fuzzy Tuned H∞ Estimator in a Low Cost AHRS

Abstract

In this paper a Fuzzy Tuned extended H∞ (FTH∞) estimator is designed and implemented in an Attitude Heading Reference System (AHRS), which is specialized to vehicular applications. In AHRS, 3-axis accelerometers are allotted to measure the earth's gravity field vector and then to update the roll and pitch angles obtained from gyros’ dynamic. Therefore, the AHRS on an accelerated vehicle will be affected by large disturbances. Additionally, in ground vehicles, the measurements of 3-axis magnetometers are corrupted by both soft and hard iron time varying disturbances. The FTH∞ estimator is an extended H∞*** filter in which only noise and attenuation bounds are tuned based on fuzzy linguistic if-then rules. The mentioned features make the estimator more reliable and suitable for hardware implementation. FTH∞ estimator relies on a consistent fuzzy combination of two change detection tests; namely, likelihood ratio and averaged norm error. Real-time implementation of new attitude-heading estimator is performed on a TMS320VC5416 Digital Signal Processor (DSP). Incorporating this powerful and small size DSP with micro electro mechanical inertial and resistive magnetic sensors leads to a low cost, small size and low power consumption AHRS. Performance of the AHRS was evaluated in Monte Carlo simulations of vehicle's attitude-headings. The FTH∞ estimator results in a superior performance compared to that of the extended H∞ estimator in simulation and real tests.