Fast calibration of a 9-DOF IMU using a 3 DOF position tracker and a semi-random motion sequence

A method for calibrating an on-body 9 Degree of Freedom (DOF) Inertial Measurement Unit (IMU) is proposed. The proposed system uses a single 3D position sensor, the earth's ambient gravity vector, the earth's magnetic field vector and a semi random motion sequence for calibration. The proposed method performs sequential calibration of each of the sensors, i.e., triaxial accelerometer, magnetometer and gyroscope, by means of two stage parameter estimation. The first stage is coarse estimation using a least squares method. The second stage is runtime estimation and smoothing of the calibration parameters using an Extended Kalman Filter (EKF). The user performs a sequence of 6 pre-defined motion primitives with the IMU during coarse estimation. The proposed method does not require exact template matching, rather it has the capability to classify and use only suitable data points for calibration. Such a calibration method is useful for applications such as home-based games and movement analysis. The results from the proposed calibration procedure were compared to those obtained using a conventional static calibration technique. The proposed calibration approach outperforms the conventional method by up to 40% in terms of calibration parameter estimation. The effectiveness of the calibration method was evaluated by estimating heading and inclination using the calibrated IMU. The results indicate that the proposed calibration procedure can allow tracking of 3D orientation with a maximum peak-to-peak error of 5°.