Previous Abstract Return to Session B10 Next Abstract

Session B10: Magnetic and Gravity Anomaly-Based Navigation 2

A Novel Loosely-Coupled Gravity-Aiding Architecture For New And Legacy Inertial Navigators
Simone B. Bortolami, Keith Sevcik, William R. Gallagher, Kenneth Ferguson, Maksim Lokhmatov, Ryan Ma, Demetri James, Armond Conte, Nicholas Saluzzi, Navigation Research and Development Center, Applied Research Laboratory, Pennsylvania State University
Location: Ballroom E
Date/Time: Wednesday, Jun. 3, 11:10 a.m.

Gravity-aided inertial navigation matches measurements of gravity anomalies (GAs) — local
deviations in Earth’s gravitational field from the standard ellipsoid model — to reference GA
maps for position fixes. GA measurements are made using dedicated accelerometers, such as in
gravimeters, or the accelerometers of the onboard inertial navigation system (INS) itself. Our
architecture can use either one.
Despite decades of research, this technology has faced persistent challenges. Chief among these
challenges is the accuracy of the maps and the nature of the field. It is widely acknowledged
that gravity-aided navigation may not be effective in all regions, as some areas lack sufficient
gravitational variations to enable accurate positioning. In addition, a detailed analysis of the
available maps reveals that map errors at higher spatial frequencies can be larger than the GAs
themselves, thereby reducing their utility for map matching. This condition especially affects
low-feature areas, while areas with high variations are less affected.
Another challenge in gravity-aided navigation is the difficulty in forming accurate GA
measurements. Gravity is measured using accelerometers whose specific force measurements
capture the GA signal along with other quantities, such as the core gravity field, vehicle platform
dynamics, and motion over the Earth. In surveying applications, these extraneous components
are separated from the GA measurement using precise GPS position and velocity. However,
gravity-aided inertial navigation doesn’t use GPS, thus the errors in indicated position, velocity,
and acceleration corrupt the GA calculations.
To address these issues, we have developed a novel architecture for gravity-aided inertial
navigation that builds upon three foundational pillars. The first of these pillars is the need to
filter both the map and the measurements to remove the error contained in the high spatial
frequencies. By filtering out high-frequency components, we utilize only those frequency ranges
where the map errors are at most commensurate with the GA measurement error. This
approach not only enhances the reliability of the map data, but also mitigates the impact of
platform motion and INS errors on the measurements. In fact, the spatial frequency of the map
errors being filtered, compounded with the cruise speed of most underwater platforms, yields
very low-pass filters that mitigate most traditional navigation errors.
The second pillar of our architecture addresses the estimation and removal of INS errors from
the GA measurement. Traditional approaches estimate INS errors and perform position fixing on
the map within a single filter solution. Our approach utilizes an INS error estimator paired with
an independent map-matching algorithm. This Kalman-type estimator, dubbed the Gravity
Passive Navigation Filter (GPNF), models the dynamics of the INS errors and removes them from
the GA measurement. The cleaned GA measurements are then passed to the third pillar of our
architecture — a multi-point fixing algorithm.
Different from traditional approaches, which recursively process measurements, our multi-point
fixing solution implements a custom terrain contour matching algorithm, or Terrain Contour
Matching (TERCOM). The TERCOM algorithm takes the corrected GA measurements provided by
the GPNF and generates position fixes by comparing them against the GA map in a closed-loop
manner. The algorithm’s multi-point design enhances stability and robustness, minimizing the
influence of individual erroneous measurements on the final position estimate. Additionally,
this approach decouples the fixing process from the INS error estimation.
The overall architecture is designed to provide loosely coupled updates to the INS, ensuring
compatibility with legacy systems and modern black-box navigators. This flexibility allows our
system to be integrated with new or existing navigation solutions, supplying corrections to the
INS in a manner similar to GPS.
In our presentation, we provide a detailed description of the architecture along with examples
of its functioning and performance. The results underscore the potential of our architecture to
augment existing navigation systems with a new GPS alternative.
This work was supported by the U.S. Navy.

Previous Abstract Return to Session B10 Next Abstract