An online recursive autocalibration of triaxial accelerometer

In this paper, we proposed a novel method for autocalibration of triaxial Micro-Electro-Mechanical systems (MEMS) accelerometer that does not require any sophisticated laboratory facilities. In particular, this method is an online calibration method which can be conveniently implemented with the accuracy of MEMS accelerometer being significantly improved. The procedure exploits the fact that the output vector of the accelerometer must match the local gravity in static state condition. To achieve online calibration, the model as well as the cost function are linearized at the beginning, and an online recursive method is then utilized to identify the unknown parameters and remove the bias caused by linearization. This online recursive method is based on damped recursive least square estimation (DRLS), which can significantly reduce the calculation complexity comparing to nonlinear optimization method. In addition, the unknown parameters can be solved in a short time and the estimated parameters can remain stable during calibration. Experimentally, this method was tested by comparing the output results before and after calibration in different condition. It showed that the output, after calibrated by the proposed method, is more accurate with respect to raw output using default factory parameters.