The Limits of In-Run Calibration of MEMS Inertial Sensors and Sensor Arrays

Abstract

MEMS accelerometers and gyroscope triads now cost less than $10, potentially opening up many new applications. However, these sensors require calibration prior to navigation use. This paper determines the maximum tolerable sensor errors for in-run calibration techniques using a basic Kalman filter by developing criteria for filter failure and performing Monte Carlo simulations for a range of different sensor specifications, and both car and UAV motion-profiles. Gyroscope bias is found to be the most significant with the maximum tolerable value of its SD varying between 0.75 and 2.6 deg/s depending on the value of the specification of the other sensor sources. The paper shows that pre-calibration and smart array techniques could potentially enable in-run calibration to be applied to lower-quality sensors. However, the estimation of scale-factor cross-coupling and gyroscope g-dependent errors could potentially be critical. Armed with this knowledge, designers can avoid both unnecessary design complexity and computational load of over-engineering and the poor navigation performance of inadequate filters. Copyright © 2016 The Authors Journal of the Institute of Navigation published by Wiley Periodicals, Inc. on behalf of Institute of Navigation.