Application of inertial sensors and flux-gate magnetometer to real-time human body motion capture

Abstract

Human body tracking for synthetic environment interface has become a significant human- computer interface challenge. There are several different types of motion capture systems currently available. Inherent problems, most resulting from the use of artificially-generated source signals, plague these systems. A proposed motion capture system is being developed at the Naval Postgraduate School which utilizes a combination of inertial sensors to overcome these difficulties. However, the current design exhibits azimuth drift errors resulting from the use of inertial sensors. This thesis proposes a new method of compensating for azimuth drift using a three-axis fluxgate magnetometer. The magnetometer capable of azimuth drift compensation since its sensitive axis is not collinear with the local vertical. This thesis includes a program for simulating the operation of a fluxgate magnetometer in C++. The included C++ code simulates a fluxgate magnetometer and provides an estimate of azimuth based on the magnetometer's output which is typically within five degrees of the actual azimuth. Real magnetometer data for testing and verification was accomplished by bench testing a real fluxgate magnetometer.