Sensor fusion based measurement of human head motion

Present virtual reality systems often use unconstrained sensors like Polhemus Tracker. These have advantages of unconstrainedness and wider motion range, but have deficiencies of low sampling rate (/spl sim/60 Hz) and critical latency (/spl sim/100 ms). This paper proposes a method, which raises the sampling rate and compensates the latency of Polhemus on rotational motion through the "sensor fusion" of gyro sensor, and Polhemus measurements and a method of evaluating the performance of unconstrained motion sensors quantatively. The validity of the methods was confirmed by off-line computation of actual data.<>