Swath bathymetry sidescan (SBS) sonar and the angle-of-arrival processing that underlies these systems has the capability to produce much higher resolution three dimensional imagery and bathymetry than traditional beamformed approaches. However, the performance of these high resolution systems is limited by signal-to-noise ratio (SNR) and they are also susceptible to multipath interference. This thesis explores two methods for increasing SNR and mitigating multipath interference for SBS systems. The first, binary coded pulse transmission and pulse compression is shown to increase the SNR and in turn provide reduced angle variance in SBS systems. The second, transmit beamforming, and more specifically steering and shading, is shown to increase both acoustic power in the water and directivity of the transmitted acoustic radiation. The transmit beamforming benefits are achieved by making use of the 8-element linear angle-of-arrival array typical in SBS sonars, but previously not utilized for transmit. Both simulations and real world SBS experiments are devised and conducted and it is shown that in practice pulse compression increases the SNR, and that transmit beamforming increases backscatter intensity and reduces the intensity of interfering multipaths. The improvement in achievable SNR and the reduction in multipath interference provided by the contributions in this thesis further strengthens the importance of SBS systems and angle-of-arrival based processing, as an alternative to beamforming, in underwater three dimensional imaging and mapping. / Graduate / 0544 / 0547 / mark.butowski@gmail.com

La classification des cibles sous-marines est principalement basée sur l'analyse de l'ombre acoustique. La nouvelle génération des sonars d'imagerie fournit une description plus précise de la rétrodiffusion de l'onde acoustique par les cibles. Par conséquent, la combinaison de l'analyse de l'ombre et de l'écho est une voie prometteuse pour améliorer la classification automatique des cibles. Quelques systèmes performants de classification automatique des cibles s'appuient sur un modèle pour faire l'apprentissage au lieu d'utiliser uniquement des réponses expérimentales ou simulées de cibles pour entraîner le classificateur. Avec une approche basée modèle, un bon niveau de performance en classification peut être obtenu si la modélisation de la réponse acoustique de la cible est suffisamment précise. La mise en œuvre de la méthode de classification a nécessité de modéliser avec précision la réponse acoustique des cibles. Le résultat de cette modélisation est un simulateur d'images sonar (SIS). Comme les sonars d'imagerie fonctionnent à haute et très haute fréquence le modèle est basé sur le lancer de rayons acoustiques. Plusieurs phénomènes sont pris en compte pour augmenter le réalisme de la réponse acoustique (les effets des trajets multiples, l'interaction avec le fond marin, la diffraction, etc.). La première phase du classificateur utilise une approche basée sur un modèle. L'information utile dans la signature acoustique de la cible est nommée « A-scan ». Dans la pratique, l'A-scan de la cible détectée est comparé à un ensemble d'A-scans générés par SIS dans les mêmes conditions opérationnelles. Ces gabarits (A-scans) sont créés en modélisant des objets manufacturés de formes simples et complexes (mines ou non mines). Cette phase intègre un module de filtrage adapté pour permettre un résultat de classification plus souple capable de fournir un degré d'appartenance en fonction du maximum de corrélation obtenu. Avec cette approche, l'ensemble d'apprentissage peut être enrichi afin d'améliorer la classification lorsque les classes sont fortement corrélées. Si la différence entre les coefficients de corrélation de l'ensemble de classes les plus probables n'est pas suffisante, le résultat est considéré ambigu. Une deuxième phase est proposée afin de distinguer ces classes en ajoutant de nouveaux descripteurs et/ou en ajoutant davantage d'A-scans dans la base d'apprentissage et ce, dans de nouvelles configurations proches des configurations ambiguës. Ce processus de classification est principalement évalué sur des données simulées et sur un jeu limité de données réelles. L'utilisation de l'A-scan a permis d'atteindre des bonnes performances de classification en mono-vue et a amélioré le résultat de classification pour certaines ambiguïtés récurrentes avec des méthodes basées uniquement sur l'analyse d'ombre. / Underwater target classification is mainly based on the analysis of the acoustic shadows. The new generation of imaging sonar provides a more accurate description of the acoustic wave scattered by the targets. Therefore, combining the analysis of shadows and echoes is a promising way to improve automated target classification. Some reliable schemes for automated target classification rely on model based learning instead of only using experimental samples of target acoustic response to train the classifier. With this approach, a good performance level in classification can be obtained if the modeling of the target acoustic response is accurate enough. The implementation of the classification method first consists in precisely modeling the acoustic response of the targets. The result of the modeling process is a simulator called SIS (Sonar Image Simulator). As imaging sonars operate at high or very high frequency the core of the model is based on acoustical ray-tracing. Several phenomena have been considered to increase the realism of the acoustic response (multi-path propagation, interaction with the surrounding seabed, edge diffraction, etc.). The first step of the classifier consists of a model-based approach. The classification method uses the highlight information of the acoustic signature of the target called « A-scan ». This method consists in comparing the A-scan of the detected target with a set of simulated A-scans generated by SIS in the same operational conditions. To train the classifier, a Template base (A-scans) is created by modeling manmade objects of simple and complex shapes (Mine Like Objects or not). It is based on matched filtering in order to allow more flexible result by introducing a degree of match related to the maximum of correlation coefficient. With this approach the training set can be extended increasingly to improve classification when classes are strongly correlated. If the difference between the correlation coefficients of the most likely classes is not sufficient the result is considered ambiguous. A second stage is proposed in order to discriminate these classes by adding new features and/or extending the initial training data set by including more A-scans in new configurations derived from the ambiguous ones. This classification process is mainly assessed on simulated side scan sonar data but also on a limited data set of real data. The use of A-scans have achieved good classification performances in a mono-view configuration and can improve the result of classification for some remaining confusions using methods only based on shadow analysis.

The space of hand gesture recognition using radar and sonar is dominated mostly by radar applications. In addition, the machine learning algorithms used by these systems are typically based on convolutional neural networks with some applications exploring the use of long short term memory networks. The goal of this study was to build and design a Sonar system that can classify hand gestures using a machine learning approach. Secondly, the study aims to compare convolutional neural networks to long short term memory networks as a means to classify hand gestures using sonar. A Doppler Sonar system was designed and built to be able to sense hand gestures. The Sonar system is a multi-static system containing one transmitter and three receivers. The sonar system can measure the Doppler frequency shifts caused by dynamic hand gestures. Since the system uses three receivers, three different Doppler frequency channels are measured. Three additional differential frequency channels are formed by computing the differences between the frequency of each of the receivers. These six channels are used as inputs to the deep learning models. Two different deep learning algorithms were used to classify the hand gestures; a Doppler biLSTM network [1] and a CNN [2]. Six basic hand gestures, two in each x- y- and z-axis, and two rotational hand gestures are recorded using both left and right hand at different distances. The gestures were also recorded using both left and right hands. Ten-Fold cross-validation is used to evaluate the networks' performance and classification accuracy. The LSTM was able to classify the six basic gestures with an accuracy of at least 96% but with the addition of the two rotational gestures, the accuracy drops to 47%. This result is acceptable since the basic gestures are more commonly used gestures than rotational gestures. The CNN was able to classify all the gestures with an accuracy of at least 98%. Additionally, The LSTM network is also able to classify separate left and right-hand gestures with an accuracy of 80% and The CNN with an accuracy of 83%. The study shows that CNN is the most widely used algorithm for hand gesture recognition as it can consistently classify gestures with various degrees of complexity. The study also shows that the LSTM network can also classify hand gestures with a high degree of accuracy. More experimentation, however, needs to be done in order to increase the complexity of recognisable gestures.

The ability of traditional active sonar processing methods to detect targets is often limited by clutter and reverberation from ocean environments. Similarly, multipath arrivals from radiating sources such as ships and submarines are received at sensors in passive sonar systems. Reverberation and multipath signals introduce constructive and destructive interference patterns in received spectrograms in both active and passive sonar applications that vary with target range and frequency. The characterization and use of interference phenomena can provide insights into environmental parameters and target movement in conjunction with standard processing methods including spectrograms and array beamforming. This thesis focuses on utilizing the time-frequency interference structure of moving targets captured on sonar arrays to enhance the resolution and abilities of conventional sonar methods to detect and localize targets. Physics-based methods for interference-based beamforming and target depth separation are presented with application of these methods shown using broadband simulated array data.

Since 1995, California State Fish Hatcheries (Feather River, Nimbus, and Mokelumne) and Coleman National Fish Hatchery have raised approximately 29 million 4 fall run Central Valley Chinook salmon (Oncorhynchus tshawytscha) per season for stock enhancement. From April through June, fish are acclimated in net-pens prior to release at one of three sites: the Carquinez Strait at Conoco Phillips (CP), the mouth of the Napa River at Mare Island (MI), and the San Joaquin River at Jersey Point (JP). Striped Bass, Marone saxatilis, are known to congregate at the release location to feed on the hatchery fish as they enter the Delta and Bay, and are suspected to be reducing numbers of Chinook recruitment. Dual-Frequency Identification Sonar (DIDSON) was used to capture video-like images to enumerate and estimate sizes of potential predators in the area. Stomach analysis was used to obtain consumption rate data and a simple model was used to estimate predator impacts on the hatchery fish. Data was collected in 2011 and 2012. In 2011 the striped bass population at CP was significantly larger than MI (p=0.009) and JP (p=0.038) and in 2011 , and MI (p=0.046) in 2012. Predators were significantly smaller (range 11.8-61.7 em, mean 34.6 em in 2011 ; 21-67 em, 42.9 in 2012) atJP (p<0.001). Average size predator at MI was 47.3 em (range 31-59 em) in 2011 and 50.9 em (range 33-73 em) in 20 12; and at CP was 48.3 em (range 16-77 em) in 2011 and 52.7 em (range 31-78 em) in 2012. On average an estimated 2.2% of hatchery fi sh are consumed each year by striped bass and predator impacts are greatest at CP (p<0.001). Changing the release site often could improve salmon survival by decreasing predator attraction to the site and reducing immediate predator-prey encounters.

No description available.

Dolphins at Kolmården dolphinarium were given a set of 20 floating fish hides, in order to simulate aforaging situation. The idea was to motivate the dolphins to use echolocation, in order to differentiatebetween hides which contained fish and hides that were empty. The dolphins would access the fish hidesfor 20-minute sessions five days per week, during a total period of five weeks. The results indicated thattheir interest in the fish hides was maintained over the entire study period for all individuals, and thisinterest did not correlate with age. However, older dolphins seemed more prone to solely inspect, possiblyby using echolocation aimed towards the fish hides than to physically interact with them. Neither was theirinterest affected by the dolphins’ pre-session activities. However, not all dolphins seemed interested in fishunless given to them by care takers. Thus, some dolphins were likely less motivated in solving theecholocation task. As a pod, the dolphins’ interest in this innovative enrichment was maintained over time,and the plasticity of these fish hides would suggest a range of different setups for the future. However,when analysing the potential use of echolocation cues, theoretical calculations of the target strengthdifferences between filled and empty fish hides, together with data on the physical interactions with them,suggested that the dolphins did not use sonar cues, but resorted to more or less random manipulation of thefish hides in order to eject the fish.

Autonomous underwater vehicles (AUVs) have the potential to dramatically improve safety, quality of life and general scientific knowledge. Our coasts, lakes and rivers are filled with various forms of marine infrastructure including dams, bridges, ship hulls, communication lines, and oil rigs. Each of these structures requires regular inspection, and current methods utilize divers, which is dangerous, expensive, and time consuming. AUVs have the potential to alleviate these difficulties and enable more regular inspection of these structures. Furthermore, there are significant scientific discoveries in the fields of geology, marine biology and medicine that AUV exploration of our oceans will enable. Since field trials of AUVs can be both expensive and high-risk, making a simulation method to generate data for algorithm development is a necessity. For this purpose, we present HoloOcean, an open-source, fully-featured, underwater robotics simulator. Built upon Unreal Engine 4 (UE4), HoloOcean comes equipped with multi-agent communications, common underwater sensors, high-fidelity graphics, and a novel sonar simulation method. Our novel sonar simulation framework is built upon an octree structure, allowing for rapid data generation and flexible usage to simulate a variety of sonars. Further, we have augmented this simulation to incorporate various probabilistic models to account for the heavy noise found in sonar imagery. Simulation enables development of many algorithms such as mapping, localization, structure from motion, controls, and many others. Localization is one essential algorithm for AUV navigation. Recent developments in the utilization of Lie Groups for robotic localization have lead to dramatic performance improvements in convergence and uncertainty characterization. One such method, the Invariant Extended Kalman Filter (InEKF), leverages that invariant error dynamics defined on matrix Lie Groups satisfy a log-linear differential equation. We lay out the various practical decisions required for the InEKF, and show that the primary sensors used in underwater robotics with minor modifications can be used in the InEKF. We show the convergence improvements of the InEKF over the quaternion-based extended Kalman filter (QEKF) on HoloOcean data, both in low and high uncertainty scenarios.

<p> In the past, analog circuits, discrete digital logic circuits or minicomputers have been used to implement the signal processing section of a sonar systems. More recently, microprocessor based logic circuit designs have produced a new breed of system design approach which gives designers the flexibility that has never been available through the use of analog or discrete logic circuits; however, due to the inherent slow speed of the metal-oxide semiconductor (MOS) logic circuits, incorporating microprocessors in the implementation of a sonar signal processor is not feasible. With the advent of bipolar Schottky large scale integrated circuit technology, the speed performance of the microprocessors have been improved considerably, and signal processor designs employing microprocessors are now feasible. </p> <p> The main objective of this work is to design, implement, and test a real-time digital sonar signal processor for processing pulsed CW signals. With design based on the use of the bit slice microprocessor, a signal processor has been constructed that has an 8 bit input, a 16 bit output. The processor is capable of detecting 16 different Doppler shifts. Laboratory generated signals are used in the testing and the experimental results show good agreement with the theory. A possible means of expanding the existing single channel signal processor into a multichannel processor has also been outlined. </p> / Thesis / Master of Engineering (MEngr)

<p> The literature on relevant parameters used in the model is reviewed. </p> <p> A Rigid Mathematical Model and a Stochastic Model are developed to describe acoustic propagation in the medium. </p> <p> The two models are used to determine bounds on the design parameters for a hypothetical shipborne SONAR System. </p> <p> The System so designed is evaluated on the basis of measurements made at sea. </p> / Thesis / Master of Engineering (MEngr)