Variable Complimentary Filter for Improving Accuracy of Human Motion Measurement

Abstract

These days, inertial sensors are often used to capture human movements. A complimentary filter is the one method of sensor fusion, which several sensors data are combined to set off each sensor's disadvantages. It is often used for wearable devices because of its high measurement accuracy at low calculation cost. Human motion contains various speeds and frequency. However, in a complementary filter, combined frequency range is fixed and the filter can not cope with motion out of the range. Thus, a coefficient of the complimentary filter is replaced with sigmoid function in order to fluctuate along with motion speed of subjects. This fluctuating coefficient enables measuring irregular motions of human. Experiment shows that using this method rises accuracy of motion measurement.