The applications of Inertial Measurement Unit (IMU) based on Micro Electro Mechanical System (MEMS) are widely used because of its low-cost, small volume and low energy consumption. Unlike the accelerometers and magnetometers, which are respectively sensitive to vibrations and ferrous materials interference, MEMS type gyroscope is able to provide stable orientation information for a short period of time. Compared with these two sensors, gyroscopes are neither disturbed by vibrations nor affected by metallic materials, and therefore it would be one of the primary sensors for industrial applications. Reviewing the most recently related works, accurate attitude estimations are usually obtained by fusing gyro integration. However, it is well known that series drifts will be induced after a long-term integration. Furthermore, according to the literature and the associated experiments, they all demonstrated that MEMS type gyroscopes have strong temperature correlation because of the silicon structure. Put it simply, the temperature variations will give rise to gyro drifts severely and thereby degrade the orientation estimation precision. To address this problem, this article proposed a temperature modeling technique and a real-time drift pre-compensation procedure by applying autoregressive moving average (ARMA) technology. Different from the traditional modeling methods, the proposed method is able to describe transient drifting behaviors as well as steady state ones. Moreover, comparison studies using artificial neural networks (ANNs) are also presented. Experiments show that the proposed method provides a stable and reliable temperature-caused gyroscope drift pre-compensation. The developed method is also very suitable for the realization in a low-cost embedded system.
All Science Journal Classification (ASJC) codes