Autocalibration of MEMS Accelerometer

In this paper, we present an autocalibration procedure for tri-axial MEMS accelerometers, which allows increasing the accuracy of this kind of micro-sensors. This makes them usable for the analysis of the human movement. The procedure is based on the constancy of the module of the accelerometer output in static condition, and it is formulated and resolved as a non linear least squares problem. The calibrated output of the MEMS compares favorably with that obtained from a traditional motion capture system. Quantitative results are reported and discussed