A minimum-order kalman filter for ambulatory real-time human body orientation tracking

In this paper, a computationally efficient orientation estimation algorithm using an inertial/magnetic sensor is presented for ambulatory real-time human motion tracking. Based on a quaternion formulation, the proposed algorithm is designed to have two main steps that are connected in feedback relationship: a quaternion measurement step with a vector selector scheme and a Kalman filter (KF) step. This allows us to choose only the quaternion as the state and measurement vectors in our KF design. Thus, the KF has a minimum-order structure (i.e., 4^th-order), which decreases the computational cost. The estimated orientation accuracy is validated experimentally by using an optical tracking system.