Theil, Stephan

Now showing 1 - 2 of 2

Some of the metrics are blocked by your
consent settings
Development and Analysis of SHEFEX-2 Hybrid Navigation System Experiment
(2013-04-29)
Steffes, Stephen
;
Holzapfel, Florian
;
Dittus, Hansjörg
;
Theil, Stephan
There is need in Europe for an accurate navigation system for Earth-based aerospace applications with high dynamics, such as launch and atmospheric entry vehicles. The Hybrid Navigation System (HNS) was developed in response to this need. This system fuses data from an IMU, a Global Position System (GPS) receiver, and a star tracker (STR) using an extended Kalman filter. Data fusion allows the user to combine the strengths of each instrument to compensate for their individual weaknesses. Combining the data from different sensors in this way reduces the effect of measurement errors from each sensor and increases navigation accuracy compared to using an uncoupled approach. This work applies modern navigation and computing methods to a real life problem and verifies these methods with flight data. Its flight on the SHEFEX2 sounding rocket mission validates the software- and hardware-in-the loop simulations, the real-time computing methods and the navigation algorithm. This is useful for future navigation missions which may base their algorithms on this work and for other electronic systems which may use these techniques for their own simulations. The novel navigation algorithm consists of a high-rate and a low-rate task running in parallel and passing data back and forth. The high-rate task processes the IMU data and integrates the strapdown navigation equations as soon as new IMU data is available (400Hz). It is also responsible for computing the linearized state transition model, calibrating clocks and correcting the whole states. The low-rate task propagates and updates the filter at 1Hz using data sent from the high-rate task and the sensors and corrects the error state vector. Together with some changes necessary for running on a real-time embedded system, this algorithm was implemented in C code and run on the HNS navigation computer. The HNS is a complex system with many devices working together in parallel. As in any mission, thoroughly testing the software and hardware is very important. To fully test the system a realistic software-in-the-loop simulator and hardware-in-the-loop test bench were developed which could simulate the expected flight conditions and test the system. Additionally, various real-world tests were done to test aspects of the system that could not otherwise be tested by simulation. These included a road test and a night sky test. SHEFEX-2 was launched on June 22, 2012 at 19:18 UTC from And{\o}ya Rocket Range in Norway by DLR. The flight results are presented here to show the performance of the individual sensors and the overall navigation system. An accurate off-line reconstruction of the trajectory is described, which is compared against the HNS flight data as an overall measure of the system's performance. The trajectory reconstruction method and navigation flight performance are explained in detail. The presented flight results show that the HNS performed very well and, except for the STR, the mission was a complete success. The SHEFEX-2 flight was the first flight of this system, which brings the HNS software and several hardware components to TRL 7 and paves the way for the HNS to be used as a mission critical subsystem in future missions.
doctoral thesis
406 345
Some of the metrics are blocked by your
consent settings
Low-cost failure-tolerant hybrid navigation designs for future space transportation systems
(2020-12-15)
Fragoso Trigo, Guilherme
;
Theil, Stephan
;
Büskens, Christof
;
Theil, Stephan
Global Navigation Satellite System (GNSS) technology is steadily making its way into space transportation systems. However, the most widely used launcher navigation systems to date, including those on board of European launch vehicles, are inertial-only. Despite reliable and robust, the dead-reckoning nature of these systems causes unbounded navigation solution drift, raising important mission constraints and limitations. Due to its bounded error characteristics, GNSS has long been used as remedy for such inertial drift. Combined, the two systems compensate many of each others’ shortcomings. This thesis presents the research and development of a robust and fault-tolerant GNSS/inertial hybrid navigation system for launchers based on lower grade sensory. A comprehensive survey of architectural elements of the GNSS/inertial fusion under the scope of the envisaged application is first carried out, providing the background and justification for the proposed system concept. The design process is broken down into three stages: baseline system concept, robust order reduction, and Fault Detection, Isolation and Recovery (FDIR) scheme design. A tightly-coupled, modular, closed-loop architecture featuring an error-state Kalman filter updated by GNSS pseudorange and time-differenced carrier phase measurements forms the baseline design. The filter runs in parallel with a strapdown inertial propagator, estimating a wide set of inertial sensor perturbations, including many associated with low-grade units, which is fed back to the propagation algorithm. GNSS receiver clock and atmospheric offsets are also mitigated within the filter. Given the large state size, filter order is then reduced through Consider state Kalman filtering. Rederiving this filtering framework, a novel perspective is drawn whereby the standard- and consider-state routines are entirely separated, enabling higher implementation flexibility. Correlation between process and measurement noise is also accounted for. Different inertial sensor filter models (high, medium, and low grade) are reduced based on extensive parametric observability and error impact analyzes using Parametric Cramér-Rao Bounds along representative launch vehicle trajectories. GNSS filter state set is also reduced. Despite using only around half of the computational load of the original full-order estimator, the resulting lower-order filter configurations maintain the robustness of the baseline design against inertial and GNSS sensor errors. Going a step further and targeting common failure modes of hybrid navigation, the developed system is suited with an FDIR module. The proposed novel scheme uses both filter innovation and GNSS measurement-set statistics to detect and isolate GNSS failures as well as inertial/strapdown faults. Threshold selection and concept validation is done through stochastic analysis of the statistical test variables under real GNSS receiver data. Required computation load is shown to be small relative to that of the filter update step. In a stride to raise the readiness of the designed system the navigation algorithm is implemented as C/C++ software, based on DLR’s Hybrid Navigation System (HNS) heritage, flight-proven code. In this process new navigation software libraries are created and several original ones updated. Despite all the additional performance, robustness, fault-tolerance enhancing features with respect to the heritage software, computational load analysis shows only a moderate burden increase, which does not considerably strain the real-time load margins of the latter. At each design stage, performance is evaluated through model-in-the-loop Monte Carlo and/or hardware-in-the-loop testing using real GNSS data from a representative receiver stimulated with trajectories of both DLR SHEFEX-2 sounding rocket and ESA Vega launcher. The C/C++ navigation software is tested in a software- and hardware-in-the-loop set-up. The proposed system is compared to several other configurations. Among these, comparison to a representative launcher (high grade) inertial-only system suggests the possibility of a one-grade reduction of the inertial sensor with the conceived hybrid design, significantly lowering system cost.
doctoral thesis
553 554

Theil, Stephan

Filters

Settings

Sort By

Results per page