Development of a Calibration Platform for Inertial Measurement Units

Abstract

Inertial Measurement Units (IMUs) are used in various industries to track the displacement and orientation of moving bodies. Interest in their use on unmanned vehicles is rising, but these sensors have one critical issue hindering their adoption: noise. What can one do about noise? Filter it. But how does one distinguish between real data and noise, and how can one select the best filter parameters?

This thesis proposes a cost-effective calibration platform for IMUs through the use of a rotating plate or swing-arm mechanism. The platform can generate a wide range of accelerations and angular velocities with a single motion, where one validates the other without using another sensor unit. Furthermore, it can create non-zero mean accelerations which are not available in commonly used shake table platforms. An experimental study is presented to validate the effectiveness of the proposed calibration platform.

The research involves designing and constructing a machine to aid in the calibration of an IMU. Data is gathered from the sensor for analysis, and data-driven and intelligent methods are used to improve filtering algorithms to reduce noise generated by machine vibrations and other environmental factors. This sensor is then be calibrated, implemented on a real-time autonomous system, and its performance is verified.