Despite significant progress in GPS/INS integration-based direct geo-referencing (DGR) technology over the past decade, its performance still needs to be improved in terms of accuracy and tolerance to GPS outages. This is mainly due to the limited geometric strength of the GPS satellite constellation, the quality of INS and the system integration technology. This research is focused on pseudolite (PL) augmentation to enhance the geometric strength of the GPS satellite constellation, and the Neural Network (NN) aided Kalman filter (KF) system integration algorithm to improve the geo-referencing system's performance during GPS outages. The main research contributions are summarized as below: a) Systematic errors introduced by pseudolites have been investigated. Theoretical and numerical analyses reveal that errors of troposphere delay modelling, differential nonlinearity and pseudolite location are sensitive to pseudolite receiver geometry. Their effect on final positioning solutions can be minimised by selecting optimal pseudolite and receiver locations, which is referred to as geometry design. Optimal geometry design for pseudolite augmented systems has been proposed based on simulation results in airborne surveying scenarios. b) Nonlinear geometry bias, or nonlinearity, exists in single difference processes when the unit vectors from the reference and user receivers to a satellite or pseudolite are non-parallel. Similar to long baseline differential GPS (DGPS), nonlinearity is a serious issue in pseudolite augmentation. A Projected Single Difference (PSD) method has been introduced to eliminate nonlinear geometry bias. An optimized expression has been derived to calculate the direction of project vectors, and the advantages of applying PSD in pseudolite augmented airborne DGPS have been demonstrated. c) A new method for pseudolite tropospheric delay modelling has been proposed, which is based on single-differenced GPS tropospheric delay models. The performance of different models has been investigated through simulations and field testing. The advantages and limitations of each method have been analysed. It is determined that the Bouska model performs relatively well in all ranges and elevations if the meteorological parameters in the models can be accurately collected. d) An adaptive pseudolite tropospheric delay modelling method has been developed to reduce modelling error by estimating meteorological parameters in real-time, using GPS and pseudolite measurements. Test results show that pseudolite tropospheric delay modelling errors can be effectively mitigated by the proposed method. e) A novel geo-referencing system based on GPS/PL/INS integration has been developed as an alternative to existing GPS/INS systems. With the inclusion of pseudolite signals to enhance availability and geometry strength of GPS signals, the continuity and precision of the GPS/INS system can be significantly improved. Flight trials have been conducted to evaluate the system performance for airborne mapping. The results show that the accuracy and reliability of the geo-referenced solution can be improved with the deployment of one or more pseudolites. f) Two KF and NN hybrid methods have been proposed to improve geo-referenced results during GPS outages. As the KF prediction diverges without measurement update, the performance of a GPS/INS integrated system degrades rapidly during GPS outages. Neural networks can overcome this limitation of KF. The first method uses NN to map vehicle manoeuvres with KF measurement in a loosely coupled GPS/INS system. In the second method, an NN is trained to map INS measurements with selected KF error states in a tightly coupled GPS/INS system when GPS signals are available. These training results can be used to modify KF time updates. Optimal input/output and NN structure have been investigated. Field tests show that the proposed hybrid methods can dramatically improve geo-referenced solutions during GPS outages.

A methodology to estimate the state of a moving marine vehicle, defined by its position, velocity and heading, from an unmanned surface vehicle (USV), also in motion, using a stereo vision-based system, is presented in this work, in support of following a target vehicle using an USV. / Includes bibliography. / Dissertation (Ph.D.)--Florida Atlantic University, 2015 / FAU Electronic Theses and Dissertations Collection

Inertial navigation systems (INSs) have the well-known advantages of being self-contained, weatherproof, jam-proof, and non-self-revealing while providing stable navigation information with little high-frequency noise. However, their single most important drawback is the growth of their error cumulatively with time in an unbounded manner. Navigation systems based on position fixing, in contrast, offer bounded errors in the long term, but their output is usually contaminated with strong high-frequency noise. To harness the advantages of both types of systems, INSs have been traditionally aided or augmented by one or more fixing system(s). Such an arrangement preserves the excellent short-term stability and damping (i.e. high-frequency rejection) capability of INSs while limiting its long-term drift. In recent years, the availability of navigation information from the Global Positioning System (GPS) reliably and accurately over the entire globe has made it a natural choice as the means of augmentation of INSs. An integrated navigation system combining data from two or more ‘pure’systems is called a hybrid navigation system (HNS). There is no unique way of combining navigation information from the INS and GPS. Depending on the goals and specifications of the overall navigation system, the instrument and equipment available, cost constraints, and technology options, the scheme for integrating INS and GPS may take one of many forms. In generic terms integration of diverse ‘pure’ navigation systems can be performed at various levels. At the simplest and most basic level, each system may be allowed to run independently, generating its own navigation data separately which may then be combined periodically to reset any accumulated error. At the other extreme, the two (or more) systems may be intimately coupled right at their raw data levels in a quasi-continuous manner with the intention of maximising their mutually beneficial effect and deriving the ‘best’ possible navigation information. Hybrid navigation architectures have been a subject of much research and development, and a significant body of information is available on the subject. However, there are clear gaps in open literature on many practical issues that arise in the context of implementing specific HNSs. In this thesis we investigate the architecture, implementation and performance issues of HNSs that combine stand-alone INS and GPS units. The thesis consists of eight chapters. The first chapter offers an introduction to the navigation problem and discusses the basic types of navigation including inertial and satellite navigation. Inertial sensors such as gyroscopes and accelerometers and the GPS are discussed in some detail. The types and principle of gyroscopes and accelerometers and the error sources in inertial navigation are briefly covered, as also the advantages and disadvantages of INS and the trends in inertial system development.The chapter also touches upon GPS segments (space, control and user), the theory and determination of position fix, and the GPS error sources. Mention is also made of the types of GPS receiver available and the trends in GPS technologies. Integration of INS and GPS and its benefits are discussed and a set of specifications for an integrated system is laid out to serve as the basis for the configurations proposed later. The second chapter, in its three sections, provides a summary of the significant literature relevant to INS and GPS in the context of their integration. Chapter three discusses mechanisation aspects of the INS-GPS hybrid navigation system. This chapter is divided into three sections. In the first section the frames of reference, INS mechanisation and the error equations are discussed. The definitions for the various frames such as body, platform, local level, geodetic, Earth-centred-Earth-fixed (ECEF), and the computer frame are mentioned along with the direction cosine matrices for the transformation of frames. In the second section various types of mechanisation of INS and the summary of tilt, velocity and position equations are described. The INS can be mechanised in two ways: the stable platform and the strap-down. In this chapter the general error equations for platform tilt, velocity and position are listed. Platform-based systems can be mechanized as one of the following types, viz. unipolar, Focualt, north pointing and wander azimuth. The definitions and summary of the tilt, velocity and position, and the error equations are given for all these types of mechanization. The accelerometer and gyro error models are discussed. It is pointed out that inclusion of all the possible INS states in the model would lead to a 45-state system which would be too complex to handle on board. The scope for reduction of model order and the effect of such reduction are brought out. The section ends with a summary of the INS error equations considered for implementation. In the third section the GPS principle and derivation of navigation solutions based on GPS measurements are dealt with. GPS error modelling, computation of DOP (dilution of precision), and clock modelling are also discussed. In this section the navigation solution for various classes of users – stationary, low-dynamics, medium-dynamics and high-dynamics – are discussed. The INS model and the clock model defined in this chapter are used in configuring the integrated system model later. Chapter four discusses the various HNS configurations and their implementation to mitigate the INS error. Three levels of integration are considered: a. Output coupled: The INS needs initial alignment during which the INS position and velocity are initialised with the precisely known co-ordinates and components at the starting location. Starting with these initial conditions, the INS-sensed accelerations are continuously integrated to yield the current velocity and position. As mentioned earlier, the INS error is dependent on this initial value and further increases with time. If the initial position and velocity inputs are precise, the short-term INS accuracy (typically for the first 10-15 minutes in case of aircraft) is usually within acceptable limits. Further error built up during longer flights can be reduced by periodic updation of INS with the precise position and/or velocity values. To achieve this the pilot may, for example, fly over waypoints whose co-ordinates are precisely known. This would constitute a physical or manual method of INS re-initialisation. A better and more modern method is to use precise GPS-derived information to reinitialise the INS periodically and automatically. b. Medium coupled: Another way of mitigating the INS error build-up is by using medium-coupled HNS wherein the INS errors are estimated using the GPS measurements as reference. The INS outputs are corrected by applying these error estimates. The important point to note here is that in medium coupling, the errors in the INS states are considered instead of the states themselves. The final geodetic outputs from the two systems are used as measurements. A twelve-state indirect feed-forward Kalman filter is used to estimate the INS position error. c. Tightly coupled: The basic measurements from the GPS are pseudoranges which are the distances from the user to the GPS satellites. By making a minimum of four such measurements the GPS receiver computes the user location in the geodetic coordinates. Conversely, knowing the user position from INS, it is possible to calculate the expected pseudoranges to known GPS satellite locations. Comparing the measured and the computed pseudoranges, the filter estimates the errors in the INS position. In this work a seventeen-state, feed-forward, indirect Kalman filter is configured to estimate the INS-derived pseudorange errors. These errors are then translated into positional errors which are used to correct the indicated INS positions. In configuring the filter it is assumed that the INS and GPS are physically separated and data transfer is through the interface buses. In this chapter the simulators used for validation and performance estimation of the configurations are also described. Two simulators are used to validate the hybrid system, namely, software-and hardware-based simulators. The simulators simulate the six-degree-of-freedom of trajectory generator, and error models of INS and GPS. The truth data from the trajectory generator are combined with the INS error and GPS error to get the INS and GPS outputs respectively. The fifth and sixth chapters covers the validation of the above-mentioned three configurations. Since analysis of output coupled systems is rather straightforward, simulation and validation of the configuration are carried out for the medium and tightly coupled systems Covariance analysis and Error analysis modes of simulation are carried out to study and validate the behaviour of the configurations. In covariance analysis one obtains the root mean square (rms) value of the errors obtained from several Monte Carlo runs. It gives an estimate of the lower bound of the system errors. Covariance simulation provides a degree of confidence in the error model but is generally not sufficient to expose the complete behaviour of the system. For detailed investigation, error simulation needs to be carried out for the entire navigation system. In the thesis, covariance simulation is carried out for both the configurations to check the sensitivity of the system to measurement update rates, process noise, update times for the transition matrix, and also for the validity of the truncation of the Taylor series expansion. The details of the simulation processes and their results are discussed in these chapter. The seventh chapter makes a performance comparison of the configurations and draws inferences for practical hybrid system implementation. From the comparisons it is seen that the loosely coupled configuration is the simplest. In this configuration there is no requirement of the Kalman filter. The accuracy depends on the update rate. If the position update is made, for example, once every 600 s then the error in the combined system will be limited to the sum of the error due to the GPS and that accumulated in the INS alone over the of 600 s interval. There is no coordinate transformation required. In the case of medium coupled filter the addition of process noise to the GPS clock model is not critical. The position accuracy achieved is around 2 m (rms). The coordinate transformations are only from the body to platform, and platform to geodetic axes. The observation matrix is very simple in this case and the computation burden is low. Dynamic tuning of the measurement matrix is not required in real time.In the case of tightly coupled configuration the addition of a certain amount of process noise deliberately to the GPS clock model is critical. The position accuracy achieved with tight coupling is around found to be 34 m (rms) without the addition of process noise. On addition of a controlled amount of noise to the GPS clock bias and clock drift states and inclusion of measurement noise as a function of GPS signal strength, the position accuracy improves significantly, to about 7m (rms). Figures 2a and 2b below depict the behaviour before and after inclusion of the noise. The coordinate transformations are from body to platform, platform to geodetic, and geodetic to ECEF coordinates, and vice versa. The observation matrix (H) for this integration model is very complicated, and the computational burden is very high. In this configuration H transfers the measurements from metres to radians. Dynamic tuning of measurement matrix is required in this case. Chapter eight of the thesis summarises the results and lists out the conclusions arrived at from the study. It also includes a section with suggestions for future work in this direction.

CAPES / Um sistema de navegação inercial (INS) é um dispositivo autônomo capaz de determinar sua própria posição a partir de medições fornecidas por sensores inerciais. Para a presente dissertação, uma ferramenta para a simulação e validação de sistemas de navegação inercial foi desenvolvida. Essa ferramenta permite que as soluções de navegação de um INS simulado possam ser comparadas a soluções de referência analiticamente exatas. A partir dos resultados dessa comparação, o usuário pode decidir pela validade ou não validade dos algoritmos de navegação do INS simulado. A ferramenta foi desenvolvida com um foco essencialmente didático para prover ao usuário um meio para a melhor compreensão do funcionamento dos complexos algoritmos associados à navegação inercial. Apesar do foco didático, a ferramenta também possui um caráter prático relevante já que ela efetivamente permite a validação de diferentes configurações de algoritmos consistentes com o estado da arte da navegação inercial. A presente dissertação, portanto, apresenta a ferramenta desenvolvida e demonstra seu correto funcionamento através de um conjunto relevante de experimentos de simulação. / An inertial navigation system (INS) is an autonomous device that determines its own position based on measurements provided by inertial sensors. For this dissertation, a simulation and validation tool for inertial navigation systems has been developed. This tool allows the navigation solutions generated by a simulated INS to be compared against analytically exact reference solutions. Based on the results of this comparison, the user can decide if the simulated INS navigation algorithms are valid or not valid. The tool has been developed with an essentially didactic focus in order to provide the user with a way to better understand how the complex inertial navigation algorithms work. Despite the didactic focus, the developed tool has also a relevant practical aspect since it effectively permits the validation of different configurations of algorithms that are consistent with the inertial navigation state of the art. This dissertation, therefore, describes the developed tool and demonstrates its correct behavior through a relevant set of simulation experiments.

Les travaux présentés dans ce mémoire de thèse concernent le développement et la mise en oeuvre d'un algorithme de filtrage particulaire pour le recalage de la navigation inertielle par mesures altimétriques. Le filtre développé, le MRPF (Mixture Regularized Particle Filter), s'appuie à la fois sur la modélisation de la densité a posteriori sous forme de mélange fini, sur le filtre particulaire régularisé ainsi que sur l'algorithme mean-shift clustering. Nous proposons également une extension du MRPF au filtre particulaire Rao-Blackwellisé appelée MRBPF (Mixture Rao-Blackwellized Particle Filter). L'objectif est de proposer un filtre adapté à la gestion des multimodalités dues aux ambiguïtés de terrain. L'utilisation des modèles de mélange fini permet d'introduire un algorithme d'échantillonnage d'importance afin de générer les particules dans les zones d'intérêt. Un second axe de recherche concerne la mise au point d'outils de contrôle d'intégrité de la solution particulaire. En nous appuyant sur la théorie de la détection de changement, nous proposons un algorithme de détection séquentielle de la divergence du filtre. Les performances du MRPF, MRBPF, et du test d'intégrité sont évaluées sur plusieurs scénarios de recalage altimétrique / This thesis deals with the development of a mixture particle filtering algorithm for inertial navigation update via radar-altimeter measurements. This particle filter, the so-called MRPF (Mixture Regularized Particle Filter), combines mixture modelling of the posterior density, the regularized particle filter and the mean-shift clustering algorithm. A version adapted to the Rao-Blackwellized particle filter, the MRBPF (Mixture Rao-Blackwellized Particle Filter), is also presented. The main goal is to design a filter well suited to multimodal densities caused by terrain amibiguity. The use of mixture models enables us to introduce an alternative importance sampling procedure aimed at proposing samples in the high likelihood regions of the state space. A second research axis is concerned with the development of particle filtering integrity monitoring tools. A novel particle filter divergence sequential detector, based on change detection theory, is presented. The performances of the MRPF, MRBPF and the divergence detector are reported on several terrain navigation scenarios

This thesis deals with inertial navigation systems issues. It describes basics of reference frames, coordinate systems and matrix calculations for AHRS. There are also basic information about inertial sensors, inertial measurements units and its mistakes. One of the purposes of this paper could be explanation of inertial navigation systems terms. The main object of this thesis is to explore the influence of using multiple sensors of same type to enhance measurements of AHRS systems.

Interferometer Fiber Optic Gyroscope (IFOG) has established as critical sensor for advance navigation systems. Sensor coil is known to be heart of IFOG. The bias drift and scale factor performance of IFOG depend on the sensor coil. The environmental perturbations like vibration, shock, temperature and magnetic field can affect the measured phase difference between the counter propagating beams, thereby introducing a bias error resulting in degradation of IFOG performance. In general these factors are both time varying and unevenly distributed throughout the coil producing a net undesirable phase shift due to variations in the optical light path, which is undistinguishable from the rotation induced signal. The development of sensor coil for high performance includes selection of optical fiber, spool material, coil winding technique and potting adhesive. In the thesis, the effects of various perturbations like temperature, vibration and magnetic field on the sensor coil are analysed, which degrades the gyro performance. The effect of temperature and vibration can be reduced by proper selection of spool material, winding method and by applying adhesive during the winding of sensor coil. The effect of magnetic field can be reduced by using the high birefringence polarization maintaining fiber with shorter beat length, shielding the sensor coil and reducing the number of twist during the winding. Design and fabrication of the sensor coil is done for control grade & navigation grade FOG with fiber length of 100 m and 1000 m respectively with the polarization maintaining fiber of two different manufacturer Fiber Core, UK and Nufern, USA selected based upon the beat length and Numerical Aperture so that sensor coil has minimum effect of magnetic field and the bending of fiber. Presently the spool material used is Aluminium alloy (HE15) for the ease of fabrication and easily availability of material. The Quadrupolar winding is done to reduce the thermal gradient effects. The indigenously developed special adhesive is applied layer by layer to reduce the environmental effects. In order to study the lifetime of sensor coil accelerated aging test (85°C, RH 85 %) for 30 days is also carried out.