Practical approach in estimating inertial navigation unit's errors

Abstract

Inertial navigation unit (INU), which is commonly composed of three orthogonally aligned accelerometers and gyros, is well known for its short term measurement accuracy in position, velocity and attitude. However, such measurement accuracy degrades with time due to various types of errors. In this paper, a practical approach is proposed to estimate both the deterministic and random errors of an INU. The deterministic errors, which include bias and scaling errors, can be estimated through a simple experimental setup; while the random noise is modeled using Allan Variance (AV) analysis method. The empirical values of the errors are then fed into the INU's system model for error correction using Kalman filtering. Finally, the calibrated INU shows promising results in preserving long term accuracy of the motion sensor.