Ubiquitous human body motion capture using micro-sensors

Traditional vision-based and commercial motion-capture systems acquire human body motion within studio. We present an ubiquitous human limb motion capture system using wearable micro-inertial-sensors (accelerometers and gyroscopes. The sensor units are attached to segments of upper limb, and the collected acceleration and angular rate data are transmitted to the PC via USB interface. Motion data are fused by a quaternion-based Kalman filter to reconstruct orientations of body segments in real time. A sensor data-driven layered human motion model is developed for real-time animation. The experimental results show that our system can capture motions and drive the animation in real-time without noticeable drift.