Optimization-based alignment for inertial navigation systems: Theory and algorithm

Inertial navigation system (INS) necessitates an alignment stage to determine the initial attitude at the very start. A novel alignment approach is devised by way of an optimization method, in contrast to the existing alignment methods, e.g., gyrocompassing and filtering techniques. This paper shows that the INS attitude alignment can be equivalently transformed into a “continuous” attitude determination problem using infinite vector observations. It reveals an interesting link between these two individual problems that has been studied in parallel for several decades. The INS alignment is heuristically established as an optimization problem of finding the minimum eigenvector. Sensitivity analysis with respect to sensor biases is made and explicit error equations are obtained for a special stationary case. Simulation studies and experiment tests favorably demonstrate its rapidness, accuracy and robustness. The proposed approach is inherently able to cope with any large angular motions, as well as high-frequency translational motions. By inspecting the constant initial Euler angles, it could alternatively be used to detect the existence of significant sensor biases.