2-Level error (drift) compensation for low-cost MEMS-based inertial measurement unit (IMU)

In this paper, we extend the 1-level compensation method for MEMS-based IMUs, dealing mainly with stochastic drifts, described in our previous work Yudanto et al. (Proc SPIE, 9517:95,172H-95---95,172H-12, 2015) to a 2-level compensation method dealing with both stochastic and temperature drifts. While the 1-level compensation method is based on an automatic detection of systems states which triggers an online recalibration of the sensors parameters, the 2-level compensation method extends the 1-level compensation method by using a model-based approach to remove the temperature influences in the sensors outputs. The 2-level compensation schemes are then integrated together to allow an algorithmic framework that can be used in an easy way to auto-calibrate MEMS-based IMUs. Experiments using an industrial MEMS-based IMU under various operating conditions have been conducted to demonstrate the added value of the 2-level compensation method. The experimental results show that the estimations of acceleration and angular velocity based on the 2-level compensation method are in general more accurate than the ones obtained based on the 1-level compensation method.