In the realm of autonomous underwater exploration and surveillance, the capability of Autonomous Underwater Vehicles (AUV) to make informed decisions in dynamic and often unpredictable environments is crucial. This thesis investigates how a software architecture can be designed and implemented for autonomous decision making on the remotely operated vehicle BlueROV2. Using raw sensor data and external information inputs, situational awareness (SA) is achieved. This is used as a basis for real-time decision making during a search mission that is set up. The thesis aims for the AUV to base its actions on the SA. This is demonstrated through a search scenario in which the AUV is assigned the task of locating a number of targets in an underwater environment. To solve the task, the AUV is able to use two different methods to solve the same problem. Either by using the sonar to locate points of interest and subsequently investigate those locations or systematically traversing the entire search area using a camera to search. The trade-off between thoroughness and time consumption makes the methods favourable in different situations. Four modules are developed and implemented in the architecture. A sensor module that collects the data, an SA module that refines the data, a decision module that handles decisions in two layers and an execution module that performs the actions. A modular architecture is developed, which enables the AUV to decide which search method to prioritise and interpret the surroundings to differentiate between targets and decoy objects. Dynamic decision-making is accomplished by changing search strategies during missions and returning when all targets are found. The targets are located with a precision high enough to make out individual targets and keep track of previously seen targets. The research contributes to the field of autonomous decision making by laying a foundation for adding more sophisticated algorithms for future problem-solving. The implemented algorithms are placeholders to showcase the capabilities developed, and the architecture is modular to enable future changes and extensions of the work.

Avhandlingen genomfördes för att ge en överblick över olika kamerasystem och möjliga algoritmer för att upptäcka och positionera EUR-pallar i ett ställage. Systemen skulle användas av autonoma gaffeltruckar. Time-of-Flight kameror, stereomoduler och structured-light kameror undersöktes tillsammans med vanliga 2D-kameror och deras objektiv. Genomförandet utvärderade två olika 2D kameror tillsammans med en ZED stereomodul från Stereolabs. Det resulterande systemet lyckades identifiera och placera en pall med mer än 5 mm precision, men resultaten var baserade på begränsad testdata.

Målet med projektet har varit att skapa en lösning som gör det möjligt för alla att färdas med självkörande bilar i framtiden, med fokus på rullstolsburna. Projektet har gjorts i nära samarbete med personer som har olika funktionsvariationer, med fokus på att kunna komma in i fordonet och spänna fast sig självständigt. Projektet har gjorts i samarbete med DRIVE-projektet på Mittuniversitet, som forskar kring energilagring, elbilsinfrastruktur och elmotordrivlina.

In this thesis we present a detect-tracking algorithm (see figure 3.1) that combines the detection robustness of static region proposal based object detectors, like the faster region convolutional neural network (R-CNN) and the region-based fully convolutional networks (R-FCN) model, with the tracking prediction strength of extended Kalman filters, by using, what we have called, a translating and non-rigid user input region of interest (RoI-) mapping. This so-called RoI-mapping maps a region, which includes the object that one is interested in tracking, to a featureless three-channeled image. The detection part of our proposed algorithm is then performed on the image that includes only the RoI features (see figure 3.2). After the detection step, our model re-maps the RoI features to the original frame, and translates the RoI to the center of the prediction. If no prediction occurs, our proposed model integrates a temporal dependence through a Kalman filter as a predictor; this filter is continuously corrected when detections do occur. To train the region proposal based object detectors that we integrate into our detect-tracking model, we used TensorFlow®’s object detection api, with a random search hyperparameter tuning, where we fine-tuned, all models from TensorFlow® slim base network classification checkpoints. The trained region proposal based object detectors used the inception V2 base network for the faster R-CNN model and the R-FCN model, while the inception V3 base network only was applied to the faster R-CNN model. This was made to compare the two base networks and their corresponding affects on the detection models. In addition to the deep learning part of this thesis, for the implementation part of our detect-tracking model, like for the extended Kalman filter, we used Python and OpenCV® . The results show that, with a stationary camera reference frame, our proposed detect-tracking algorithm, combined with region proposal based object detectors on images of size 414 × 740 × 3, can detect and track a small object in real-time, like a tennis ball, moving along a horizontal trajectory with an average velocity v ≈ 50 km/h at a distance d = 25 m, with a combined detect-tracking frequency of about 13 to 14 Hz. The largest measured state error between the actual state and the predicted state from the Kalman filter, at the aforementioned horizontal velocity, have been measured to be a maximum of 10-15 pixels, see table 5.1, but in certain frames where many detections occur this error has been shown to be much smaller (3-5 pixels). Additionally, our combined detect-tracking model has also been shown to be able to handle obstacles and two learnable features that overlap, thanks to the integrated extended Kalman filter. Lastly, our detect-tracking model also was applied on a set of infra-red images, where the goal was to detect and track a moving truck moving along a semi-horizontal path. Our results show that a faster R-CNN inception V2 model was able to extract features from a sequence of infra-red frames, and that our proposed RoI-mapping method worked relatively well at detecting only one truck in a short test-sequence (see figure 5.22).

Made available in DSpace on 2016-04-25T20:20:10Z (GMT). No. of bitstreams: 1 Edgar da Nobrega Gomes.pdf: 824162 bytes, checksum: 18786700ef4232606428a633b6dbb4a5 (MD5) Previous issue date: 2011-05-06 / In a time of crisis, of biases and paradoxes, there are strong evidences regarding the inadequacies of the hegemonic development model. At the same time, there is no global evidence that an alternative way is possible. Worldwide changes with complex relations affect business, families and governments. Economic, political, social, environmental issues until recently, were analyzed based on indicators that were efficient before but no longer make sense. In this uncertain time, plenty of questions are made, but there are no answers strong enough to move "hearts and minds . Is it possible to construct other forms of organization of society? Can solidarity be considered a fundamental element of economic ideas? The construction of a theory that helps steer a new path for humanity will be the result of new values and new assumptions. Therefore, we must overcome the differences that are, sometimes, more "appearance that essence". Considering the limits and possibilities of practical experience and theoretical analysis of Solidarity Economy, there is elements that reaffirm this as an alternative utopian realism. The conclusion of this study was that there are two important issues for an agenda that considers solidarity an essential paradigm for a new economy: freedom and participation. These are central elements, the result of of conscious choices that individuals make and use these results participating rightly and wrongly, without fear of moving forward or back. It is the model of cooperative or voluntary work, solidarity economy, where people can fully exercise their freedom. Participation becomes increasingly a necessity of development processes / Em um tempo de crise, de polarizações e paradoxos, há fortes evidências em relação às insuficiências do modelo de desenvolvimento hegemônico e, ao mesmo tempo, não se evidencia globalmente a possibilidade de uma alternativa. As transformações que afetam o planeta com suas complexas relações impactam de maneira multifacetada as empresas, as famílias e os governos. Aspectos econômicos, políticos, sociais e ambientais até há bem pouco tempo eram analisados a partir de indicadores que apresentavam respostas que hoje não fazem mais sentido. Nesses tempos de incertezas, uma série de questionamentos é colocada sem que existam respostas fortes o suficiente para mover corações e mentes . É possível construir outras formas de organização da vida em sociedade? A solidariedade pode ser considerada elemento fundamental das concepções econômicas? A construção de uma teoria que ajude a orientar um novo caminho para a humanidade será resultante de novos valores e novos pressupostos. Portanto, é preciso ultrapassar as diferenças que às vezes são mais aparência que essência . Diante dos limites e das possibilidades das experiências concretas e das análises teóricas da Economia Solidária há elementos que reafirmam esta como uma alternativa de realismo utópico. Neste trabalho concluiu-se que existem duas questões importantes para uma agenda que considera a solidariedade um paradigma essencial para uma nova economia: liberdade e participação. Estes são elementos centrais e não são resultantes apenas de desejos, mas de escolhas conscientes que os indivíduos fazem e que as exercitam participando, acertando e errando, sem medo de seguir em frente ou voltar atrás. É no modelo cooperativo ou associativo de trabalho, na Economia Solidária, que as pessoas podem exercer na plenitude a sua liberdade. A participação passa a ser cada vez mais uma necessidade dos processos de desenvolvimento

Made available in DSpace on 2016-04-26T14:53:00Z (GMT). No. of bitstreams: 1 Edgar da Nobrega Gomes.pdf: 824162 bytes, checksum: 18786700ef4232606428a633b6dbb4a5 (MD5) Previous issue date: 2011-05-06 / In a time of crisis, of biases and paradoxes, there are strong evidences regarding the inadequacies of the hegemonic development model. At the same time, there is no global evidence that an alternative way is possible. Worldwide changes with complex relations affect business, families and governments. Economic, political, social, environmental issues until recently, were analyzed based on indicators that were efficient before but no longer make sense. In this uncertain time, plenty of questions are made, but there are no answers strong enough to move "hearts and minds . Is it possible to construct other forms of organization of society? Can solidarity be considered a fundamental element of economic ideas? The construction of a theory that helps steer a new path for humanity will be the result of new values and new assumptions. Therefore, we must overcome the differences that are, sometimes, more "appearance that essence". Considering the limits and possibilities of practical experience and theoretical analysis of Solidarity Economy, there is elements that reaffirm this as an alternative utopian realism. The conclusion of this study was that there are two important issues for an agenda that considers solidarity an essential paradigm for a new economy: freedom and participation. These are central elements, the result of of conscious choices that individuals make and use these results participating rightly and wrongly, without fear of moving forward or back. It is the model of cooperative or voluntary work, solidarity economy, where people can fully exercise their freedom. Participation becomes increasingly a necessity of development processes / Em um tempo de crise, de polarizações e paradoxos, há fortes evidências em relação às insuficiências do modelo de desenvolvimento hegemônico e, ao mesmo tempo, não se evidencia globalmente a possibilidade de uma alternativa. As transformações que afetam o planeta com suas complexas relações impactam de maneira multifacetada as empresas, as famílias e os governos. Aspectos econômicos, políticos, sociais e ambientais até há bem pouco tempo eram analisados a partir de indicadores que apresentavam respostas que hoje não fazem mais sentido. Nesses tempos de incertezas, uma série de questionamentos é colocada sem que existam respostas fortes o suficiente para mover corações e mentes . É possível construir outras formas de organização da vida em sociedade? A solidariedade pode ser considerada elemento fundamental das concepções econômicas? A construção de uma teoria que ajude a orientar um novo caminho para a humanidade será resultante de novos valores e novos pressupostos. Portanto, é preciso ultrapassar as diferenças que às vezes são mais aparência que essência . Diante dos limites e das possibilidades das experiências concretas e das análises teóricas da Economia Solidária há elementos que reafirmam esta como uma alternativa de realismo utópico. Neste trabalho concluiu-se que existem duas questões importantes para uma agenda que considera a solidariedade um paradigma essencial para uma nova economia: liberdade e participação. Estes são elementos centrais e não são resultantes apenas de desejos, mas de escolhas conscientes que os indivíduos fazem e que as exercitam participando, acertando e errando, sem medo de seguir em frente ou voltar atrás. É no modelo cooperativo ou associativo de trabalho, na Economia Solidária, que as pessoas podem exercer na plenitude a sua liberdade. A participação passa a ser cada vez mais uma necessidade dos processos de desenvolvimento

<p>I en obemannad flygfarkost måste datorer ta över pilotens förmåga att värdera risker och undvika kollisioner. På algoritmnivå brukar man dela in problemet i tre delar: Upptäckt och estimering av inblandade farkosters positioner och hastigheter, kollisionsriskberäkning och slutligen undanmanöver.</p><p>Saabs arbete med obemannade farkoster har tidigare berört kollisionsundvikning lite ytligt men nu börjat på större allvar. Det här examensarbetet är en del i denna satsning och har resulterat i ett sätt att beräkna kollisionsrisken samt ett sätt att beräkna en undanmanöver, givet att de inblandade farkosternas positioner och hastigheter är kända. </p><p>I examensarbetet behandlas parvisa kollisionsscenarier mellan ickekommunicerande farkoster givet två olika fall. Dels där den främmande farkostens position skattats väl, dels där den främmande farkostens position skattats sämre. En enkel simuleringsmiljö har utvecklats, där två algoritmer för beräknandet av kollisionsrisken, en för varje fall, testats samtidigt som undanmanövern testats för en mängd kollisionsscenarier. Givet att den främmande farkostens position skattats väl behöver den obemannade farkosten cirka 6 s på sig för att kunna undvika en kollision. I fallet där den främmande farkostens position skattats sämre kan vi beräkna kollisionsrisken och i vissa fall sluta oss till hur farkosterna är orienterade och därigenom göra ett undanmanöverval.</p> / <p>Saabs work with unmanned aerial vehicles has only scratched the surface of collision avoidance, but is now advancing. This master thesis sheds light on some parts of the collision avoidance problem and has resulted in an innovative way to calculate the risk of collision and a way to determine an avoidance maneuver.</p><p>In this master thesis collision scenarios between non-communicating vehicles are being looked upon in pairs, given two different sets of data. Good estimates of the unknown vehicles position and unsatisfying position estimate. Through the development of a simple simulation environment, two algorithms, one for each set of data, has been tested simultaneously with tests of the collision avoidance maneuver for several collision scenarios. Given a good estimate of the unknown aerial vehicles position, the unmanned vehicle need approximately 6 seconds to act to avoid a collision. For the case with unsatisfactory estimate of the unknown vehicle the risk of collision can be calculated and in some cases the orientation of the aerial vehicles and thus a choice of avoidance maneuver can be made.</p>

I en obemannad flygfarkost måste datorer ta över pilotens förmåga att värdera risker och undvika kollisioner. På algoritmnivå brukar man dela in problemet i tre delar: Upptäckt och estimering av inblandade farkosters positioner och hastigheter, kollisionsriskberäkning och slutligen undanmanöver. Saabs arbete med obemannade farkoster har tidigare berört kollisionsundvikning lite ytligt men nu börjat på större allvar. Det här examensarbetet är en del i denna satsning och har resulterat i ett sätt att beräkna kollisionsrisken samt ett sätt att beräkna en undanmanöver, givet att de inblandade farkosternas positioner och hastigheter är kända. I examensarbetet behandlas parvisa kollisionsscenarier mellan ickekommunicerande farkoster givet två olika fall. Dels där den främmande farkostens position skattats väl, dels där den främmande farkostens position skattats sämre. En enkel simuleringsmiljö har utvecklats, där två algoritmer för beräknandet av kollisionsrisken, en för varje fall, testats samtidigt som undanmanövern testats för en mängd kollisionsscenarier. Givet att den främmande farkostens position skattats väl behöver den obemannade farkosten cirka 6 s på sig för att kunna undvika en kollision. I fallet där den främmande farkostens position skattats sämre kan vi beräkna kollisionsrisken och i vissa fall sluta oss till hur farkosterna är orienterade och därigenom göra ett undanmanöverval. / Saabs work with unmanned aerial vehicles has only scratched the surface of collision avoidance, but is now advancing. This master thesis sheds light on some parts of the collision avoidance problem and has resulted in an innovative way to calculate the risk of collision and a way to determine an avoidance maneuver. In this master thesis collision scenarios between non-communicating vehicles are being looked upon in pairs, given two different sets of data. Good estimates of the unknown vehicles position and unsatisfying position estimate. Through the development of a simple simulation environment, two algorithms, one for each set of data, has been tested simultaneously with tests of the collision avoidance maneuver for several collision scenarios. Given a good estimate of the unknown aerial vehicles position, the unmanned vehicle need approximately 6 seconds to act to avoid a collision. For the case with unsatisfactory estimate of the unknown vehicle the risk of collision can be calculated and in some cases the orientation of the aerial vehicles and thus a choice of avoidance maneuver can be made.

<p>Denna rapport beskriver styrbeteenden för autonoma agenter och utvärderar hur användbar den tekniken är för att skapa flexibla agenter i kontinuerliga miljöer. Arbetet undersöker hur väl det går att använda styrbeteenden i en mer strikt kontrollerad miljö, nämligen biltrafik. De utvecklade agenterna agerar i en trafikmiljö där de navigerar för att följa en väg och undvika kollisioner med andra agenter. Agenterna använder en kombineringsarkitektur som gör det möjligt att använda olika beräkningsmodeller. Agenterna är testade i två scenarion där teknikens användbarhet och flexibilitet är studerad. Resultatet från utvärderingen visar att tekniken är väldigt flexibel men kan kräva mycket finjustering för att uppnå bästa resultat.</p>

Thesis (D. Min.)--Boston University, 1999. / Abstract. Includes bibliographical references (leaves 241-252).