Transform-Domain Mechanization for Software-Based Enhancement of Inertial Navigation Performance

QuNav proposes to develop a TRansform-domain INS Mechanization and Algorithmic Correction (TRIMAC) in order to improve the accuracy of inertial navigation in the absence of external aids. TRIMAC will implement a transform-domain INS mechanization, which uses an exact analytical solution of underlying inertial differential equations (as opposed to traditionally applied numerical approximations) thus eliminating computational errors (such as conning and sculling). When no external navigation aids are available, mitigation of measurement errors (due to IMU sensor errors such as noise, biases, scale-factors, etc.) will be addressed using a probabilistic motion model. A stochastic motion model will be developed for missile applications. A partitioned estimator will be then applied for joint model identification/INS error estimation. Specifically, a non-linear Bayesian estimator will be used to identify critical disturbance modes. The maximum likelihood mode (or a subset of modes) will be then applied for the error estimation and subsequent drift mitigation using near-linear batch estimation techniques. Batch estimation is particularly beneficial for a) decoupling of INS error states and unknown modeling parameters via null-space projection; and, b) robust estimation in presence of non-Gaussian outliers via statistical gating and probabilistic data associating filtering. Approved for Public Release | 16-MDA-8917 (15 November 16)