The Performance Evaluation of Gravity and Zero Velocity Measurement Based Field Calibration Methods Applicable for Various Grades of Inertial Sensors

Inertial sensors basically composed of accelerometers and gyroscopes are widely used as a crucial part of the navigation systems for various types of civil and military applications. Depending upon the design technologies and manufacturing procedures, inertial sensors contain different types of systematic and stochastic errors in their measurements. The systematic error sources such as bias effect, scaling error, cross-coupling effect of non-sensitive sensor axis measurement etc. result in navigation frame acceleration error, which is double integrated to yield unbounded positioning error to inertial navigated system. In order to reduce the systematic errors built in those sensors, extensive laboratory calibration procedures are well developed with the usage of precision testing equipment such as two or more axis rotation tables, flight motion simulators and precision dividing head equipment. Since the laboratory calibration procedures are essentially depend on rigorous human source, lengthy procedures and highly expensive precision equipment, the idea of inertial sensor calibration without use of precision and costly equipment is adopted in this paper. Based on the previous literature information about the usage of local gravity and zero-velocity measurements for static conditions as a physical quantity is exploited and Extended Kalman Filter based calibration algorithm is developed. Both simulation analysis and real life experiments imply that simple rotation sequences with non-accurate mechanical devices or even with manually via human hand can be used to dictate each of three axis of inertial measurement unit sensor's errors to become observable states. Simulation analysis provides that both of the local gravity measurement and zero-velocity measurement information can be used individually to calibrate both of the accelerometers triad and gyroscopes triad deterministic error terms including repeatable bias errors, scaling errors, cross-coupling errors and g-sensitive bias errors. In this paper, the developed calibration algorithms are tested with near-navigation, tactical and commercial grade inertial measurement units and the performance goal is achieved by comparison with laboratory calibration results.