ESA GNC Conference Papers Repository
Architectural elements of hybrid navigation systems for future space transportation
Navigation on-board of the European launcher families is, to this day, performed in a purely inertial manner. Measurements from strapped-down accelerometers and gyroscopes are integrated providing a high-rate navigation solution. Despite being robust, the uncorrected inertial integration unavoidably drifts over time. Acceptable propagation error growth is only achieved with highly accurate sensors, which are generally very costly and heavy. Moreover, the (still) drifting navigation solution poses important mission performance and operational constraints. Injection accuracy, for instance, highly depends on navigation solution quality and, consequently, deteriorates with mission length. Nominal mission trajectory and profile are also highly dependent on inertial drift, as large enough flight boundaries need to be accommodated in the already strict map of no-fly areas and radar visibility regions. The result is often far from optimal. GNSS measurements are often combined with inertial ones, in what is known as GNSS-INS (hybrid) navigation, as a remedy for drift in land-based and aeronautical applications. The low-rate satellite-based information corrects the high-rate inertial propagation while inertial data smoothly bridges GNSS outages. In a launch environment, however, GNSS technology faces important vulnerabilities. Being a non-self-contained system, external signal disturbances and disruptions can occur (e.g. jamming, spoofing, tropospheric and ionospheric effects). Additionally, the receiver tracking loops are not immune to the high-dynamics, vibration and shocks of launch flight. Although ever present, these risks and effects can potentially be reduced by the combination with inertial measurements. This paper looks into several key design elements of a GNSS-INS hybrid navigation system applied to space transportation. The most fundamental system architectural options such as coupling philosophy or depth (e.g. loose, tight, ultra-tight), modularity between filter and inertial integration, and open-/closed-loop nature of the configuration, are here discussed and traded-off in the light of the envisaged application. System performance with different inertial sensor grades, from cheaper MEMS-class to more expensive navigation-grade, in different degrees of integration complexity is also looked into. Given that lower grade sensory carries higher magnitude and diversity of measurement errors and noises, especially in such a dynamically demanding environment, possible ways to circumvent this issue are investigated. A GNSS receiver outputs a variety of measurements, from processed navigation solutions (position, velocity) to raw quantities (pseudoranges, pseudorange-rates, carrier phases), and, for some receivers, even smoothed versions of these. As a result, for a given coupling depth a choice of measurement set used to update the filter must be made. This selection is here analysed for loosely and tightly coupled configurations. The error and noise profiles of such measurement quantities under launch dynamics are also looked into. This allows not only for a more educated selection, but also supports the modelling of these observables within the navigation filter. Modelling options for error sources such as receiver clock bias and tropospheric delay are discussed and compared. Finally, given the strict reliability and availability requirements of launcher navigation, we discuss the robustness and fault-tolerance of hybrid navigation configurations, from robustness-added filter designs to active fault-detection, isolation and recovery algorithms. Acknowledgements This work is part of the ESA NPI funded 'Low-Cost Failure-Tolerant Hybrid Navigation Designs for Future Space Transportation Systems' project, contract number 4000111837/14/NL/MH, under Technical Officer S. Bennani.