Session C2, Paper #1

Avionics-Based Integrity Augmentation System for Mission-and Safety-Critical GNSS Applications
R. Sabatini, Cranfield University, UK; T. Moore, C. Hill, University of Nottingham, UK

Space Based and Ground Based Augmentation Systems (SBAS/GBAS) have been developed in the US, in Europe and in other countries to improve GNSS integrity, accuracy and availability for aircraft navigation and landing applications. Along with SBAS and GBAS, GNSS augmentation may take the form of additional information being provided by other avionics systems. In most cases, the additional avionics systems operate via separate principles than the GNSS and, therefore, are not subject to the same sources of error or interference. A system such as this is referred to as an Aircraft Based Augmentation System (ABAS). The additional sensors used in ABAS may include Inertial Navigation Systems (INS), TACAN/VOR-DME, Radar, Vision Based Sensors, etc. Unlike SBAS and GBAS technology, research on ABAS is limited and mainly concentrates on additional information being blended into the position calculation to increase accuracy and/or continuity of the integrated navigation solutions. Additionally, no significant attempts have been made of developing ABAS architectures capable of generating integrity signals suitable for safety-critical GNSS applications (e.g., aircraft precision approach and landing) and no flight certified ABAS products are available at present. During flight test activities with GNSS and Differential GNSS (DGNSS) systems, it was observed that one or more of the following conditions was prone to cause navigation data outages or severe performance degradations:  Antenna obscuration due to aircraft manoeuvring;  Bad satellite geometries and low SNR values;  Doppler shifts caused by aircraft-satellites relative motion;  Interference, at the airborne GNSS antenna, caused by non-GNSS RF signals;  Multipath caused by GNSS signals reflected by the Earth's surface or the aircraft body. The last two problems can be mitigated by existing technology solutions (i.e., filtering of RF signals, proper antenna installation and software masks). However, there is little to nothing one can do in order to prevent critical events during realistic test/training manoeuvres and particular approach procedures (e.g., curved approaches) performed with high performance military aircraft. Furthermore, although in some cases a careful mission planning may significantly reduce the number of GNSS outages, the adoption of specific aircraft piloting strategies (using the information currently available in the cockpit) cannot effectively avoid the occurrence of these events. This paper presents the research activities carried out by the Italian Air Force Flight Test Centre in collaboration with Nottingham Geospatial Institute and Cranfield University in the area of GNSS Avionics Based Integrity Augmentation (ABIA). The research included design, integration and experimental flight test activities carried out on MB-339CD, TORNADO and TYPHOON aircraft. As soon as the validity of the ABIA concept was established, a prototype system was developed for use in flight test applications. This system is capable of alerting the pilot when the critical conditions for GPS signal loss are likely to occur (within a specified maximum time-to-alert). In this ABIA prototype, the aircraft on-board sensors provide information on the aircraft relevant flight parameters (navigation data, engine settings, etc.) to an Integrity Flag Generator (IFG), which is also connected to the on-board GPS receiver. The IFG can be incorporated into one of the existing airborne computers or can be a dedicated processing unit. Using the available data on GPS and the aircraft flight parameters, integrity signals are generated which are displayed on one of the cockpit displays and sent to an Aural Warning Generator. At the same time, an alternate flight path is computed taking into account the geometry and the tracking status of the available GPS satellites, together with the current mission requirements and the information provided by the aircraft Flight Test Instrumentation (FTI) and standard on-board sensors. Current research is extending the results obtained from flight tests to the design of an Advanced Avionics Based Integrity Augmentation System (AIAS) suitable for manned and unmanned aircraft applications. Mathematical algorithms have been developed to cope with the main causes of GNSS signal outages and degradation in flight, namely: obscuration, multipath, fading due to adverse geometry and Doppler shift. Adopting these algorithms, the AIAS system is able to provide steering information to the pilot and electronic commands to the aircraft/UAV flight control system, allowing real-time avoidance of safety-critical flight conditions and fast recovery of the required navigation performance in case of GNSS data losses. This is achieved by implementing both caution (predictive) and warning (reactive) integrity flags, as well as 4-Dimensional Trajectory (4DT) optimisation models suitable for all phases of flight (in line with SESAR and NextGen operational requirements). In other word, the AIAS system is able to address both the predictive and reactive nature of GNSS integrity augmentation, and implements 4DT optimisation algorithms that will streamline the system global acceptance and its introduction in the next generation airspace. The detailed design of the AIAS IFG module was completed and validation activities were performed on TORNADO-IDS and A-320 simulated platforms to determine the Time-to-Alert (TTA) performances of the AIAS system in various flight phases, including GBAS assisted precision approach and landing. The results of these activities were very encouraging, showing that the system TTA performances are in line with current ICAO, FAA and RTCA requirements for the different phases of flight, including CAT-III precision approach and auto-landing. Current research is concentrating on the 4DT optimisation module design and is extending the scope of the IFG validation to UAV platforms, investigating the potential of the AIAS system in other mission- and safety-critical GNSS applications. The initial results of this research are showing that the AIAS system could provide an significant step-forward in achieving a certifiable UAV Sense-and-Avoid (SAA) capability both in cooperative applications (e.g., ADS-B) and in non-cooperative scenarios (e.g., integration of GNSS with vision-based and other sensors). In conclusion, although current and likely future SBAS/GBAS augmentation systems can provide significant improvement of GNSS navigation performance, it is anticipated that the ABIA system will play a key role in GNSS integrity augmentation for mission- and safety-critical applications such as aircraft precision approach/auto-landing and UAV SAA. Furthermore, using suitable data link and data processing technologies, a certified ABIA capability could be a core element of a future GNSS Space-Ground-Avionics Augmentation Network (SGAAN).

[Return to Program]

Session C2, Paper #2

Interacting Multiple Model and Probabilistic Data Association Filter on Radar Tracking for ATM System
Y-C. Kao, S-S. Jan, National Cheng Kung University, Taiwan

According to the International Civil Aviation Organization's (ICAO) plan, the communications, navigation, surveillance, and air traffic management (CNS/ATM) system based on the Global Navigation Satellite System (GNSS) technology is implemented to replace the traditional air traffic control (ATC) system which is mainly based on the ground-based radar. Due to the low signal power of GNSS, it is a major concern that the CNS/ATM system may experience service interruption when GNSS signal is blocked by either intentional or unintentional radio frequency interference. With the aim to maintain the normal operation of the CNS/ATM system, one of the promising approaches is to use the existing ground-based radar system to provide navigation and surveillance services for ATM system. In order to utilize the existing radar system as a backup solution to GNSS, the tracking capability of the radar system has to be enhanced to be compatible with the standard GNSS positioning services. As a result, the objective of this paper is to implement a tracking algorithm which improves the aircraft tracking performance of the radar system under various flight modes. The tracking algorithm implemented in this paper is called the interacting multiple model (IMM) estimator which provides tracking estimates with significant noise reduction and fast response to sequences of aircraft maneuver modes. Another problem of the radar system is to track an aircraft correctly in the cluttered environment, because there might be more than one tracking observations for a single aircraft under such environment, that is, there are some tracking measurements which are not originated from the target aircraft among all the returning measurements. Therefore, this paper uses the probabilistic data association (PDA) filter to assign all of the validated measurements with different weights. The PDA filter obtains an estimator which incorporates all the returning measurements that might originate from the target of interest rather than select only one of them, and the resulting estimator makes a validation gate to determine whether the measurements would be used. Next, the validated measurements are combined with different weights due to their locations, and then use the combined measurement to update the state estimate of the target. Consequently, the PDA filter can extend the tracking capability into the region of high cluttered environment and has computational requirements suitable for real time application. Accordingly, this paper first introduces the algorithms of the IMM estimator and the PDA filter. The integration of the IMM estimator and the PDA filter are investigated in this paper as well, and the integrated filter is called the IMMPDA filter (IMMPDAF). This IMMPDAF has the capability to correctly track an aircraft under different maneuver modes and track an aircraft in the cluttered environment with adequate tracking performance. The implementation procedure of the IMMPDAF to a radar system is then discussed. The tracking performance of the IMMPDAF is compared with that of the filter comprised by the Nearest Neighbor (NN) method as well as that of the standard Kalman filter. Additionally, the computation loads of these filters will be evaluated as well. The simulation results indicate that the IMMPDAF outperforms the other filters with acceptable increasing computation load. Finally, the real radar data collected by the Civil Aeronautics Administration (CAA) of Taiwan is used to demonstrate the improvement on tacking accuracy of the IMMPDAF.

[Return to Program]

Session C2, Paper #3

GPS-INS integration for Aircraft Automatic Landing with Altimeter and Vision Sensors
S. Agarwal, Cranfield University, UK ; V.V. Unhelkar, IIT Bombay, India; A. Sharma, IIT Kanpur, India; H.B. Hablani, IIT Bombay, India

In this paper, we study the performance of a Global Positioning System (GPS) receiver integrated with a Micro Electro-Mechanical Sensor (MEMS) grade inertial navigation system (INS) and altimeter/vision sensors for landing of unmanned aircraft in urban environments, and other applications. Varying in complexity and computational requirements, different approaches of integration of GPS and altimeter/vision sensors with INS are studied. Simulations are developed to examine the performance of the algorithms. The main focus of this work is to examine the effect of aiding INS with vision sensors, altimeter and GPS, in GPS-challenged environments. To enhance the autonomy of aerial vehicles in situations where access to global localization information is not guaranteed, a system which combines monocular vision with INS measurements in Extended Kalman Filter (EKF) framework is proposed. This enables accurate, drift-free estimation of the vehicle attitude even when GPS signals are lost. Additionally, aiding from altimeter measurements is included to enhance the achievable accuracy and robustness. This approach is compared with the conventional approach of combining raw pseudorange measurements of the GPS receiver with the INS using a single Extended Kalman Filter for navigation, without any additional augmentation. To achieve further precision in navigation, estimation of bias states of the accelerometer and gyros, and use of dual frequency receivers is also examined. Standard single frequency Global Positioning System (GPS) receivers provide a positioning accuracy of approximately 4-20 m. This precision is further enhanced with dual frequency receivers which are able to provide accuracy of around 1-12 m. However, these errors are large for critical applications such as aircraft landing. Carrier phase tracking is a differential-GPS approach which allows range determination with a higher level of accuracy. However, carrier phase measurements require estimate of unknown fixed integer ambiguities before the receiver can start determining its position. Using single-difference smoothed pseudorange measurements the integer ambiguities can be estimated with reasonable accuracy, resulting in further reduction in the position error. Thus, a modified integration algorithm, which includes measurements from dual-frequency GPS receivers, is also analyzed. Choice of the algorithm for a particular application can be made based on the type of available hardware, computational capability and desired accuracy. In order to examine the performance of the approaches described above, simulations of aircraft landing incorporating the above navigation algorithms are carried out. The navigation filter estimates the position, velocity, attitude and additional states related with sensor errors. The INS is modeled with the errors in accelerometers and gyroscopes considered as time-varying biases and white noises. The GPS is modeled by incorporating models of satellite clock errors, receiver clock bias, tropospheric delay, ionospheric delay and receiver measurement errors. To determine the closed-loop performance of the system, a controller is also included in the simulation. The longitudinal mode of the aircraft in landing configuration is examined. The autopilot is designed to minimize the vertical deviation from the designed trajectory while maintaining its airspeed. Variations of the Extended Kalman Filter based GPS-INS integration, using the techniques and augmentations described above, i.e., with and without (i) bias estimation, (ii) dual frequency receiver, and (iii) vision/altimeter augmentation, are presented for aircraft navigation. Further, performance of these methods for the critical phase of aircraft landing is examined and compared through simulations. Although performance of these algorithms has been mainly examined for aircraft landing, they can be applied for any aerial vehicles as well as for extended period of flights. Robustness of the system with altimeter/vision augmentation is studied by simulating the navigation filter with intermittent GPS outages. The results of the simulations indicate that augmenting the GPS-INS filter with vision/altimeter suppresses the drift inherent in a MEMS IMU, and so, the proposed augmentation is a viable alternative for autonomous navigation of aerial vehicles in GPS-challenged environments.

[Return to Program]

Session C2, Paper #4

Integrated GNSS/INS Positioning Combined with SRTM Digital Elevation Model for Safe Approach, Landing and Takeoff Phase of a Flight
A. Ciecko, University of Warmia and Mazury in Olsztyn, Poland

The paper presents preliminary results of the project concerning integration of GNSS/INS (Global Navigation Satellite System/Inertial Navigation System) solution with SRTM (Shuttle Radar Topography Mission) DEM (Digital Elevation Model) for air navigation. The project has been financed by National Science Centre of Republic of Poland. The project assumes creation of a mathematical model of air navigation based on GNSS and INS solution as well as integration of the result with a digital elevation model (DEM). Such a solution will allow the future implementation of a fully functional navigation system in any type of the aircraft. Thanks to the system, the pilot will have a permanent access to an accurate and reliable information on the actual elevation and possible obstacles in the vicinity of the plane in each phase of flight, regardless of weather conditions. It is estimated that about one-third of serious accidents involving aircraft transport is the result of hitting the Earth's surface by a plane performing completely smooth and normal flight plan, run by well-prepared and efficient staff, these accidents are commonly called CFIT - Controlled Flight Into Terrain. In April 2010, a Polish Air Force Tupolev Tu-154M aircraft crashed in Smolensk (Russia), killing 96 people including President of Poland and his wife in CFIT accident. Disasters are often caused by errors of pilots, but their incidence is always composed by many additional factors among which a central role play harsh weather conditions and unexpected equipment failures. Innovative application proposed under this project, which combines the best features of GNSS/INS would increase the possibility of avoiding the tragedy in these situations. The paper presents in detail theoretical and practical possibilities of Septentrio AsteRx2i HDC - a dual-frequency GPS/GLONASS/INS receiver purchased within the project. Description of algorithms used in the receiver's software as well as results of first practical experiments with Septentrio AsteRx2i HDC are given . The first trials were done with a car which allowed for acquaintance with the receiver and practical tests of different configuration of measurement unit. Preliminary results of the flight trials are also presented in the paper. Detailed results of preliminary tests, advantages and disadvantages as well as theoretical possibilities of AsteRx2i HDC are presented and discussed. An accurate and reliable position determined by integrated GNSS/INS technology will be an ideal input for mathematical model which will allow for safe air navigation. Proposed solution will use the latest achievements and technologies, including the latest localization technologies to increase integrity and safety during critical phases of a flight, especially during the approach, landing and takeoff of an aircraft. The paper presents currently used integration strategies for GNSS and INS: loose, tight, ultra-tight and deep integration and an attempt of best method selection for the critical air navigation operations. The review concerns possibility of use of presented strategies with Septentrio AsteRx2i HDC receiver. Proposed model of safe air navigation also assumes prediction of airplane positioning based on previous trajectory. Discussion on possible prediction models is also presented. Current status of the project as well as planned practical experiments and future actions will sum up the paper.

[Return to Program]

Session C2, Paper #5

Multi-constellation GNSS/INS to Support LPV-200 Approaches and Autolanding
M. Orejas, J. Dunik, Z. Kana, Honeywell International, Czech Republic

Reliable navigation information is a critical aspect of aviation, especially when more and more aircraft are to occupy the same volume of airspace. The level of required accuracy and integrity depends on the phase of flight that an aircraft needs to undertake. GPS performance has been continuously improving in the past decades; however, standalone GPS is still not sufficient to support approaches with vertical guidance or autolanding. It is anticipated that future GNSS scenario, in particular GPS modernization and Galileo deployment, may provide enough performance to support these operations. For the last couple of years, a special focus of research has been the assessment of the capability of future multi-constellation airborne receivers to support LPV-200 approaches in order to offer an alternative to current Space Based Augmentation Systems (SBASs). Nevertheless, even if the performance proves to meet the requirements, GNSS will still be susceptible to various threats, e.g. interference, spoofing, scintillation, which can result in failures that directly impact the continuity of the service. One of the most effective ways to overcome this limitation is the integration with an inertial navigation system (INS). In the event of a failure a hybrid GNSS/INS system will limit the degradation of the performance for certain period of time, compared to a standalone GNSS, increasing the probability of the navigation system to complete the targeted procedure. For how long the navigation system will be able to meet the specified requirements will depend on the type of failure and the grade of the inertial sensors utilized. In aviation, the INS is typically based on navigation grade sensors. However, for procedures that require relatively short time for completion, e.g. approaches with vertical guidance and autolanding, utilization of hybrid systems with lower grade sensors are of particular interest. This paper investigates the robustness against different threats and improvement in continuity of service by using different grades of inertial sensors in a hybrid multi-constellation GNSS/INS system (a dual-frequency GPS/Galileo receiver is assumed). A tightly coupled GNSS/INS system, in feed-back (or closed loop) configuration with an error state based Extended Kalman Filter, is presented and used to quantify continuity and integrity performance. The pseudorange error is modeled as carrier noise (it is assumed smoothing is implemented) plus a pseudorange bias. This bias is modeled as a first order Gauss-Markov process and is incorporated by the filter for each satellite being processed. The steady-state covariance of the process is determined by sum of the covariances of each error contributing to the total User Equivalent Range Error (UERE): ionospheric (in case of single frequency mode), tropospheric, multipath, receiver noise, and signal-in-space (SiS) errors. Each of these errors have different correlation times, thus, a unified time constant based on a least observability analysis was chosen. This was done through Monte-Carlo simulations and selecting the time constant yielding minimum position accuracy. For the clock error, typical TCXO clock bias and drift were modeled; the parameters used were extracted from their Allan variance parameters. The inertial sensor errors modeled comprise: white measurement noise, constant and dynamic bias, scale factor, and misalignment. The integrity scheme used for the hybrid GNSS/INS was based on the solution separation method generalized for a Kalman-filter based implementation. The original method was modified to compute both, horizontal and vertical, protections levels (HPL and VPL) and to take into account the occurrence of multiple single faults and constellation faults. Furthermore, an algorithm that attempts to jointly optimize the probability of false alarm and miss detection for each sub-solution was developed in order to minimize the computed protection levels. Two sets of experiments were performed to assess the benefits of the developed hybrid navigation system: simulation of LPV-200 approaches and simulation of autolanding. In both cases different grades of inertial sensors and various failures were simulated. The list of failure scenarios include: total loss of GNSS, e.g. due to scintillation, loss of several satellites, e.g. due to excessive masking, switch to single frequency mode, e.g. due to interference in L5 band, complete loss of one constellation. For each combination of failure scenario and inertial sensor, Monte-Carlo simulations were performed. The first set of experiments presented evaluates the ability of the hybrid GNSS-INS system to support an LPV-200 approach procedure when a failure occurs at the final approach fix and continues until reaching the decision height (DH). From the output of the simulations, the most stringent requirement for each case was identified and the average time of service availability, i.e. period of time in which all specified requirements are met, was computed. The second set of experiments presented in the paper focus on the assessment of autopilot performance with the developed navigation system in the loop. An autolanding procedure aims to perform autonomous landing maneuver of aircraft. The autopilot should met the requirement of allowable touchdown dispersion for the normal case at a 10^-6 probability level. A Gaussian distribution of the total system error was assumed in order to compute the required standard deviation of the touchdown error for the longitudinal and lateral axes. The autolanding simulation model consists of the aircraft model and the airborne navigation system. The aircraft model contains true data validated dynamic model of business aircraft and its autopilot (inner loop controlling the engine thrust and aero surface deflections and outer loop controlling speed and attitude). For this set of simulations the failure occurs at the DH and last until touchdown. Results illustrate the relation between the vertical navigation and the horizontal landing error and the impact of the inertial sensors' grade on the autolanding performance.

[Return to Program]

Session C2, Paper #6

Flight Test Evaluation of eDME Performance Enhancements
K. Li and W. Pelgrum, Ohio University

Since its inception in 1952, the Distance Measuring Equipment (DME) has formed an essential part of aviation navigation. An extensive DME ground infrastructure is present globally, and most aircraft are outfitted with DME interrogator equipment. The transition to Performance Based Navigation (PBN) [1] increases the performance demands on the navigation systems, which warrants modernization of the existing DME/N system to enhanced DME or eDME. The eDME system is targeted to improve the accuracy, availability, continuity and integrity of the current DME/N system while maintaining backward compatible with existing interrogators. In [2][3], a highly-accurate measurement setup is introduced that enables the performance evaluation of a future enhanced DME system using current legacy transponder installations. The data collection system provides precisely time-stamped RF recordings of both the DME pulses as transmitted by the transponder and of the pulses received by the aircraft antenna. Post-processing of these data recordings, combined with accurate timing and truth, allow for a detailed analysis of various eDME concepts. Several flight tests have been conducted with this system by Ohio University Avionics Engineering Center in 2011 and 2012, and have indicated significant potential for eDME performance enhancements. The concept of eDME carrier phase tracking was first introduced in [2]. DME carrier phase measurements allow for millimeter-level displacement measurements, which opens a wide range of eDME performance improvements. This paper further details the implementation and performance of eDME carrier phase, as well as some potential applications and their flight-test performance evaluation. Carrier phase measurements are obtained using Phase Locked Loop (PLL) tracking of the continuous carrier phase of all received DME pulses. The paper compares these tracking results with their theoretical optima. Pulse pseudorange measurements are calculated by differencing the measured Time Of Arrival (TOA) and the measured Time of Transmission (TOT) of every transmitted DME pulse. The "Carrier-Smoothed Pulse pseudorange"-algorithm (CSP) is applied to significantly reduce the noise and high-frequency multipath on the measured pseudoranges. "Pulse pseudorange Minus Carrier" (PMC) is used to identify multipath on the pulse-ranging. Observing the peak-to-peak Pulse minus Carrier over at least one period of the slowest multipath fading frequency (typically the ground multipath) effectively overbounds the multipath over that observation interval, thereby enabling assured ranging performance. With the Pulse-Noise-Multipath algorithm, a period with minimal multipath is selected based on the PMC multipath overbound. During that period, the bias between the carrier phase and pulse pseudorange is determined, after which the calibrated carrier phase is used for ranging. Re-calibration of the carrier phase bias is performed every time the PMC algorithm indicates a lower multipath level than during the previous calibration, or after a configurable time period or displacement, whichever comes first. The paper presents extensive flight test results of the abovementioned algorithms under a variety of flight conditions. References: [1] International Civil Aviation Organization (ICAO) "Performance-based Navigation (PBN) Manual," ICAO Doc 9613, 3rd edition, 2008, ISBN 978-92-9231-198-8 [2] K. Li, W. Pelgrum, "Optimal Time-of-Arrival estimation for Enhanced DME," Proceedings of the 24th International Technical Meeting of The Satellite Division of the Institute of Navigation (ION GNSS 2011), Portland, OR, September 2011 [3] K. Li, W. Pelgrum, "Flight Test Performance of Enhanced DME (eDME)", Proceedings of the 2012 International Technical Meeting of The Institute of Navigation, Newport Beach, CA, January 2012.

[Return to Program]

Session C2, Paper #7

eDME Architecture Development and Flight-Test Evaluation
W. Pelgrum, K. Li, M. Smearcheck, Ohio University

Distance Measuring Equipment, DME, plays a crucial role in the current and future aviation navigation. DME has good performance, wide-spread international coverage, decades of proven robustness, and dissimilar failure modes from satellite-based navigation systems. These characteristics reaffirm DME's potential in current and future aviation Positioning Navigation and Timing (PNT). However, with the transition to Performance-Based Navigation (PBN) [1] the requirements for PNT get significantly more stringent. This warrants upgrades of the DME/N system to enhanced DME, or eDME. This paper presents various potential eDME architectures, proof-of-concept implementations, and flight-test results. Substantial performance enhancements can be obtained by evolutionary technology advancements. An increased number of transponders, improved transponder timing and processing, increased transponder capacity, modern interrogator technology: combined they significantly improve the DME system performance level beyond the currently specified [2] 0.2 nmi total system ranging error. An essential next step is the revision of the DME/N performance standards such that full credit can be taken for this improved performance. Further DME performance improvements can be obtained by more "revolutionary" enhancements. The eDME concept is typically associated with the addition of a UTC-synchronized "beat" signal broadcasted by the transponders [3], which will enable passive ranging (pseudoranging) and hence unlimited capacity, as well as the provision of time to the user. Data broadcast is generally also considered part of the eDME enhancements. A dramatic performance improvement both in terms of accuracy and integrity can be achieved by DME carrier phase tracking [4]. Carrier phase tracking combined with pseudoranging enables a range of algorithms such as Pulse pseudorange Minus Carrier, Carrier smoothed Pulse pseudoranging and Pulse-Noise-Multipath, which increases accuracy and simultaneously provides assurance for this performance. Robust, accurate, assured, and cost-effective time synchronization of a large number of DME transponder sites is not trivial. This paper presents various options ranging from robust GNSS-based time transfer, 2-way time transfer, one-way time transfer using geostationary satellites, and time distribution using LEO satellites. Even without transponder synchronization eDME beat-signal broadcasts have significant merit, assuming that the transponder is outfitted with an ultra-stable oscillator (for example a Rubidium-ensemble or Cesium oscillator.) This paper presents a combination of pseudoranging with occasional two-way ranging that allows the receiver/interrogator to resolve the timing offsets between transponders. This dramatically improves the capacity of DME without the challenge of transponder time synchronization. Flight test results indicate the potential of this concept. References: [1] International Civil Aviation Organization (ICAO) "Performance-based Navigation (PBN) Manual," ICAO Doc 9613, 3rd edition, 2008, ISBN 978-92-9231-198-8 [2] Department of Transportation, Federal Aviation Administration, "Performance Specification Distance Measuring Equipment (DME)," FAA-E-2996, April 1, 2008 [3] L. Eldredge, et al., "Alternative Positioning, Navigation & Timing (PNT) Study," International Civil Aviation Organization Navigation Systems Panel NSP), Working Group Meetings, Montreal, Canada, May 2010 [4] K. Li, W. Pelgrum, "Flight Test Performance of Enhanced DME (eDME)", Proceedings of the 2012 International Technical Meeting of The Institute of Navigation, Newport Beach, CA, January 2012.

[Return to Program]

Session C2, Paper #8

Potentials of Inertial Navigation Aided by Position Broadcasts and Relative Measurements
H. Mokhtarzadeh and D. Gebre-Egziabher, University of Minnesota, Twin Cities

This paper examines the feasibility of cooperative navigation using relative range measurements and information sharing as a means to limit inertial navigation system (INS) error growth in the absence of absolute measurements. This parallels a scenario of ADS-B/TCAS being used by commercial aircraft as a backup navigation system during a GPS outage. This work presents an availability study and proposes a suitable distributed estimator architecture for the problem. The availability study computed aircraft densities near representative US airports using commercial flight trajectory data from 2005. The definition used for availability under the cooperative navigation application is the fraction of time three of more aircraft, referred to as collaborators, are within the communication range of the ownship aircraft. A minimum of three collaborators is the theoretical minimum required to unambiguously estimate the position of ownship. This probability was found to be sensitive to the location, time of day, and horizontal cooperation range. Little sensitivity was observed to vertical cooperation altitude. Availabilities between 30?80% were common at major airports like LGA and MSP in an area with a 30 nmi radius around the airport during morning to evening hours. While insufficient if considered exclusively, cooperative navigation has the potential to act as part of the complete alternative positioning, navigation, and timing (A-PNT) solution in the case of a GPS outage. The on-board INS aided by cooperative navigation would serve to reduce the error growth rate and thereby enable longer periods of operation under nominal separations. A sub-optimal decentralized estimator architecture applicable to cooperative navigation serving as a backup during a GPS outage is identified. The Sequentially Partitioned Algorithm (SPA) is used to develop the estimator architecture[1]. Divergence can be mitigated by judiciously selecting which member of the cooperative navigation network are allowed to share or exchange with each other. This is known as the source selection problem and has been used for similar applications in the past, notably the Relative Navigation (RelNav) function of Joint Tactical Information Distribution System (JTIDS) (as described in the open literature)[2,3]. The estimator described here is different from the JTIDS-RelNav filter and is designed to take into account the computational and communication bandwidth constraints that exist with ADS-B, and TCAS avionics. Design issues associated with information exchange between cooperating vehicles in a distributed navigation network (e.g., communication bandwidth, number of cooperating agents, error correlation, etc) are discussed. Approaches used to deal with them in the context of the estimator developed for this work are also discussed. In particular, the issue associated with correlated errors and their effect on the stability of the cooperative navigation solution are highlighted. For example, in most cooperative navigation schemes, the attractive computation and practical features of the decentralized estimator are partly achieved by dropping states accounting for inter-vehicle error correlations. If left unrestricted, this could lead to filter divergence[4]. Source selection schemes like fixed-rank and covariance based source selection are compared and shortcomings of each for the described application are identified. The introduction of source selection improves filter stability, but the SPA estimator treatment of collaborator uncertainty as independent additive noise causes the estimate covariance to become overly confident. The improper covariance estimates are unacceptable for stringent integrity requirements. To deal with this a novel approach is presented whereby the estimator architecture is modified to take advantage of Covariance Intersection (CI), an existing conservative technique for fusing multifilter estimates with unknown correlations[5]. Finally, preliminary results based on a SPA filter using fixed-rank source-selection + CI to aid the on-board INS with relative range measurements are presented. References: [1] Titli, A., and Singh, M. G.: Systems: Decomposition, Optimization and Control, Pergamon Press, 1978 [2] Fried, W. R., and Loeliger, R.: Principles, System Configuration and Algorithm Design of the Inertially Aided JTIDS Relative Navigation Function, Navigation: Journal of The Institute of Navigation 26(3), 224-236, 1979 [3] Kerr, Thomas H., and Chin, Leonard: A Stable Decentralized Filtering Implementation for JTIDS Relnav, IEEE PLANS, 1980 [4] Gobbini, G. F.: Relative Navigation by means of Passive Ranging, Massachusetts Institute of Technology, 1981 [5] Julier, S. J., and Uhlmann, J. K.: A Non-divergent Estimation Algorithm in the Presence of Unknown Correlations, Proceedings of the American Control Conference, June 1997.

[Return to Program]

Session C2, Alternate #1

Flight Test Evaluation of Predictive Rendering Image Navigation for Close-Formation Flight
S. Calhoun, J. Curro, J. Raquet, Air Force Institute of Technology

This paper describes the flight test evaluation of an image-based relative navigation system for close-formation aerial refueling applications. The system under test relies on an Electro-Optical (EO) camera and a set of Embedded Inertial Global Position Systems (EGI) to determine a precise relative position vector between a KC-135R tanker and a Lear jet receiver aircraft surrogate. The system was developed by the Advanced Navigation Technology (ANT) Center at the Air Force Institute of Technology (AFIT) in support of the Air Force Research Laboratory's Automated Aerial Refueling (AAR) program. The AAR program is developing various technologies and procedures for the purpose of demonstrating automated refueling concepts for the USAF manned and unmanned aircraft. Flight test execution was based around standard aerial refueling profiles, focusing mainly on operations near the tanker (<2km). The main objective related to the EO system was to collect precisely time-tagged imagery and inertial data to perform a post-flight feasibility evaluation, focusing mainly on relative navigation accuracy and service robustness. The predictive rendering image navigation approach is based upon using correlation techniques between actual sensor imagery and a 3-Dimensional (3-D) model to generate 2-Dimensional (2-D) pseudo-images to solve for the relative position with respect to the tanker. The estimated relative pose is determined by perturbing the relative position state and generating the pseudo-images based on the tanker and camera models until a maximum correspondence, using standard image correspondence techniques such as Sum Squared Difference (SSD), is achieved. To mitigate the effects of lighting conditions, edge detection and other feature detection schemes are applied to both the imagery and pseudo-imagery prior to the correspondence step. Accuracy of this process is assessed by using high-precision Novatel GPS receivers on both the tanker and Lear jet to obtain a relative truth vector. Results presented from the flight tests show very good performance, supporting the claim that this approach is a potential candidate for precise formation flight and aerial refueling applications, at least from an accuracy perspective. The paper includes discussions on some challenges encountered as part of the post-test analysis. This includes processing time and, depending on the correspondence approach, convergence on local minimums and/or maximums. Additionally, some interesting aspects of the flight test build-up are discussed in the paper. Camera calibration and bore-sighting of the EO sensors, which proved to be critical to achieving the necessary accuracy, are covered. Much of the current published calibration work related to EO-based navigation applications focus mainly on solving for and understanding the various intrinsic camera parameters associated with the sensor and lens components such as distortion parameters, principal points, focal lengths and the like. However, as discovered through this testing, the more challenging aspect for this testing was related to the boresighting of the system and solving for the translational and rotational parameters to align the sensor to the aircraft body. Procedures for solving for these parameters using a survey grade laser theodolite and some lessons learned are briefly discussed.

[Return to Program]

Session C2, Alternate #2

Automated Aerial Refueling Using Scanning LiDAR
J. Curro and J. Raquet, Air Force Institute of Technology

This paper focuses on using a scanning Light Detection and Ranging (LiDAR) to determine a relative position solution for two aircraft performing aerial refueling. The relative position solution enables the two aircraft to control themselves in order to maintain a relative position to allow Autonomous Aerial Refueling (AAR). This effort is motivated by the introduction of Unmanned Aerial Vehicles (UAVs) into the United States Air Force (USAF) which require the capability for aerial refueling.Current solutions to the AAR problem use the Global Positioning System (GPS) While very accurate, the GPS approach can have problems with satellite acquisition when the tanker aircraft blocks the view of the sky for the receiver aircraft. Also, GPS signals are very low power that are subject to electronic interference. AFRL is specifically looking into different sensors that can complement GPS for AAR. One such sensor is a LiDAR.LiDAR has typically been used in navigation to determine the surrounding environment. The LiDAR provides information about the unpredictable environment, such as object tracking and plane detection, to aid in navigation. However, typical flight navigation is in a predictable environment without many other objects in close proximity. This allows navigation without a high accuracy position estimate. In these cases, a LiDAR can be used to navigate by scanning the ground. A notable exception to these cases is the aerial refueling situation where the environment will contain at least one other plane in close proximity. A LiDAR sensor now can be useful for flight navigation to determine the location of the other plane. This allows the LiDAR to search specifically for one object instead of searching for any number of unknown objects. Because of this assumption, navigation with a LiDAR in flight has many advantages over normal LiDAR navigation, and can achieve a more accurate position solution.This paper focuses on using the LiDAR to determine an accurate relative position solution between a receiver aircraft and a tanker. The attitude of each aircraft is provided by high quality Inertial Navigation Systems (INS). The aircraft are assumed to start in close proximity within range of the LiDAR with an accurate estimate of the initial relative position. This paper analyzes two different algorithms to determine relative position. One method attempts to fit the LiDAR measurement to a known model of the tanker aircraft using the Iterative Closest Point Algorithm (ICP). The Model Based ICP (MBI) algorithm determines the closest point on the estimated position of the tanker model to the current measurements and fits the two point clouds together to calculate a position estimate. The second algorithm called the Position Perturbations Difference (PPD) algorithm predicts LiDAR measurements based on different search positions and compares the estimated measurements to the actual measurements. The position with the best comparison metric is chosen as the location of the tanker.These algorithms were tested using both simulation and real flight data. Flight data was collected from and AFRL-led test involving a KC-135 tanker and a Calspan Learjet with a custom-built nosecone which contained a LiDAR, an inertial measurement unit (IMU), and a camera. Only the LiDAR and IMU data was used for the results presented in this paper. The position estimates using the two algorithms demonstrated using a scanning LiDAR approach to determine relative position during aerial refueling is possible. Relative position estimates using simulated data had 4.3cm Mean Radial Spherical Error (MRSE) for the MBI algorithm and 12.4cm MRSE for the PPD algorithm. Relative position estimates using actual data had 33.9cm MRSE for the MBI algorithm and 39.1cm MRSE for the PPD algorithm after modeling corrections were applied. Root Mean Square (RMS) error in the forward-back axis of the tanker body frame is larger compared to other axes while using the flight test LiDAR setup. Position estimates using simulated data with the Custom LiDAR setup decreased the RMS error in the forward-back axis but increased the vertical axis RMS error compared to the position estimates using simulated data with the flight test LiDAR setup. Also, The MRSE for the Custom LiDAR setup was lowered to 2.99cm MRSE compared to the 4.3cm MRSE of the flight test LiDAR setup. This shows the LiDAR setup can shift RMS error from one axis to another or reduce the MRSE altogether.Attitude and position estimates were simulated using a modified MBI algorithm with the Custom LiDAR setup and showed Euler angle RMS errors under 1 Milli Radian (MRAD) and MRSE of 2.7cm. The flight test LiDAR calculated attitude and position estimates with the simulated data using the modified MBI algorithm as well, and achieved Euler angle RMS errors of about 3MRAD with a MRSE of 6.7cm. Attitude and position estimates on actual flight test data were poor and had an MRSE of 4.3m. The PPD + MBI algorithm however, obtained a MRSE of 52.7cm and Euler angle RMS errors under 25MRAD when attitude and position estimates were calculated from actual flight test data. This shows that position estimates can be calculated even if the attitude is not provided by the IMUs.Based on the results, either algorithm has an accuracy of about 40cm MRSE for real flight test data after applying modeling corrections. Based upon comparing simulated and real results, we think that the final errors are a result of a combination of modeling error and errors in lever arms and orientations. With more precise lever arms and orientations (and perhaps a more accurate model), the MRSE can be further reduced to achieve more accurate position estimates. Attitude and position estimates can be achieved with the PPD + MBI algorithm and result in MRSE of about 52cm with Euler angle RMS errors under 25MRAD.

[Return to Program]