Numerical Differentiation of Tracking Data of Human Motion: The Virtual Accelerometer
Abstract
A method based on the Kalman Filter for deriving whole-body segment accelerations from position tracking data is described. Such a procedure is experimentally calibrated and qualified as a Virtual Accelerometer supplying a resolution of 0.2 m/s2 and a relative accuracy of better than 10 percent rms within 0–8 Hz bandwidth. Results are finally compared to accelerometer performances reported by previous investigators.