Modified strapdown inertial navigator error models

The paper revisits the problem of error modeling for strapdown INS for the purpose of navigation sensor blending with a Kalman filter. This problem has been addressed repeatedly over the last 30 years, and different perspectives on INS error modeling have emerged. The paper reviews these, focussing in particular on the properties and relative advantages and disadvantages of the φ-angle and ψ-angle error models. The paper then addresses some new concepts for the design of a Kalman filter model for integrated navigation. The sometimes troublesome problem of the explicit occurrence of the specific force in both error models is averted with the modified φ-angle and ψ-angle error models proposed in the paper, in which the explicit representation of the specific force is canceled via a transformation of the velocity error states