Autocalibration of Triaxial MEMS Accelerometers With Automatic Sensor Model Selection

Up to now, little attention has been posed on a principled derivation of the cost function used for autocalibration of MEMS tri-axial accelerometers. By formulating the calibration problem in the context of maximum likelihood estimate, we derive here a general formulation that can be reduced to the classical quadratic cost function under certain hypotheses. Moreover, we adopt the Akaike information criterion to automatically choose the most adequate linear sensor model for the given calibration data set. Experiments on simulated and real data show the effectiveness of the proposed approach.