Effectiveness of model-based motion estimation from an inertial measurement unit

This article examines the effectiveness of using a mathematical model to improve motion estimates from an inertial measurement unit (IMU). Our custom built IMU (termed WIMOTIONZ) is capable of measuring acceleration, angular rate, and the magnetic field vector in three axes. A magnetometer calibration technique that removes external magnetic disturbances is discussed, and the objective function used in the optimisation routine is presented. The addition of a model is shown to decrease the RMS error, with respect to a “gold-standard” encoder, by 67% compared with only using a magnetometer. Combining the IMU and model is shown to predict the position of the sensor to within 1% of its physically measured value.