京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第18966号 / 工博第4008号 / 新制||工||1617(附属図書館) / 31917 / 京都大学大学院工学研究科都市社会工学専攻 / (主査)教授 小池 克明, 教授 石田 毅, 准教授 水戸 義忠 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第18969号 / 工博第4011号 / 新制||工||1618(附属図書館) / 31920 / 京都大学大学院工学研究科都市社会工学専攻 / (主査)教授 小池 克明, 教授 松岡 俊文, 教授 田村 正行 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

Recent research makes use of biologically inspired computation and artificial intelligence as efficient means to solve real-world problems. Humans show a significant performance in extracting and interpreting visual information. In the cases where visual data is not available, or, for example, if it fails to provide comprehensive information due to occlusions, tactile exploration assists in the interpretation and better understanding of the environment. This cooperation between human senses can serve as an inspiration to embed a higher level of intelligence in computational models. In the context of this research, in the first step, computational models of visual attention are explored to determine salient regions on the surface of objects. Two different approaches are proposed. The first approach takes advantage of a series of contributing features in guiding human visual attention, namely color, contrast, curvature, edge, entropy, intensity, orientation, and symmetry are efficiently integrated to identify salient features on the surface of 3D objects. This model of visual attention also learns to adaptively weight each feature based on ground-truth data to ensure a better compatibility with human visual exploration capabilities. The second approach uses a deep Convolutional Neural Network (CNN) for feature extraction from images collected from 3D objects and formulates saliency as a fusion map of regions where the CNN looks at, while classifying the object based on their geometrical and semantic characteristics. The main difference between the outcomes of the two algorithms is that the first approach results in saliencies spread over the surface of the objects while the second approach highlights one or two regions with concentrated saliency. Therefore, the first approach is an appropriate simulation of visual exploration of objects, while the second approach successfully simulates the eye fixation locations on objects. In the second step, the first computational model of visual attention is used to determine scattered salient points on the surface of objects based on which simplified versions of 3D object models preserving the important visual characteristics of objects are constructed. Subsequently, the thesis focuses on the topic of tactile object recognition, leveraging the proposed model of visual attention. Beyond the sensor technologies which are instrumental in ensuring data quality, biological models can also assist in guiding the placement of sensors and support various selective data sampling strategies that allow exploring an object’s surface faster. Therefore, the possibility to guide the acquisition of tactile data based on the identified visually salient features is tested and validated in this research. Different object exploration and data processing approaches were used to identify the most promising solution. Our experiments confirm the effectiveness of computational models of visual attention as a guide for data selection for both simplifying 3D representation of objects as well as enhancing tactile object recognition. In particular, the current research demonstrates that: (1) the simplified representation of objects by preserving visually salient characteristics shows a better compatibility with human visual capabilities compared to uniformly simplified models, and (2) tactile data acquired based on salient visual features are more informative about the objects’ characteristics and can be employed in tactile object manipulation and recognition scenarios. In the last section, the thesis addresses the issue of transfer of learning from vision to touch. Inspired from biological studies that attest similarities between the processing of visual and tactile stimuli in human brain, the thesis studies the possibility of transfer of learning from vision to touch using deep learning architectures and proposes a hybrid CNN that handles both visual and tactile object recognition.

No description available.

No description available.

[pt] A estabilidade de encostas constituídas por solos residuais não saturados sofre grande influência das variações das condições de fluxo locais, uma vez que, em muitos casos, grande parte de sua resistência ao cisalhamento é conferida pela coesão aparente acarretada pelo fenômeno de sucção. Sabendo disso, o presente trabalho avaliou, através de análises de estabilidade e de fluxo realizadas em modelos bi e tridimensionais, a possibilidade de se criar um modelo onde o resultado final de estabilidade independa da distribuição de sucção inicial. As análises e modelagens de fluxo e estabilidade foram executadas com auxílio dos softwares SVDesigner, SVSlope e SVFlux-GE integrantes do pacote SVoffice, desenvolvido pela SoilVision. Buscou-se quantificar a influência da distribuição da sucção inicial nos casos estudados. Os parâmetros de solo utilizados como dados de entrada nos modelos derivaram de estudos desenvolvidos por de Oliveira (2013) sobre os mecanismos de ruptura de movimentos de massa que ocorreram em Nova Friburgo durante o megadesastre que ocorreu na Região Serrana do Rio de Janeiro em 2011, e os dados climáticos advieram de registros contemporâneos e pretéritos ao evento. Nos modelos 2D foi possível, ao se considerar aproximadamente oito meses de dados climáticos, que diferentes condições iniciais de sucção homogêneas convergissem para um resultado único de estabilidade. Nas modelagens também foi observado que o comportamento do fluxo diferiu entre os modelos 2D e 3D o que teve um significativo impacto em suas análises de estabilidade. / [en] The stability of slopes consisting of unsaturated residual soils is strongly influenced by variations in local flow conditions, since in many cases much of their shear strength is conferred by the apparent cohesion caused by the suction phenomenon. With this in mind, the present study evaluated the possibility of creating a model where the final stability result is independent of the initial suction distribution through stability and flow analyzes performed in two and three-dimensional models. Flow and stability analyzes and modeling were performed using the SVDesigner, SVSlope and SVFlux-GE software components of the SVoffice package developed by SoilVision. It was tried to quantify the influence of the initial suction distribution in the cases studied. The soil parameters used as input data in the models were derived from studies developed by de Oliveira (2013) on the mechanisms of rupture of the mass movements that occurred in Nova Friburgo during the disaster that occurred in the Serrana Region of Rio de Janeiro in 2011, and the climatic data came from records in the region during and before the event. In the 2D models, it was possible, considering approximately eight months of climatic data, that different initial homogeneous suction conditions converged to a single stability result. It was also observed that the flow behavior differed between the 2D and 3D models, which had a significant impact on their stability analyzes.

My most recent body of work explores the synthesis of my influences, interests and life experiences into imagery of common themes: The expression of dynamic figures and forms and colors in digital 3d space, cinematic composition, and vibrant color, expressed through a semblance of Aztec culture and wildlife. My sculptures of nature and ancient culture are created using contemporary digital art creation technologies and techniques. I examine the art and religion of the Aztecs and the universal search for understanding and purpose in the world and the forces around and beyond us.

Assigned sex estimation via the greater sciatic notch (GSN) is traditionally performed via physical/visual examination and ordinal scoring; however, this relies on the subjective assessment of morphology for typological classification which may not be reflective of human variation. Three-dimensional (3D) photogrammetry may offer a technologically advanced, low cost, and more objective alternative to assess the complex curvature of anatomical landmarks. This research explores the accuracy of photogrammetry derived 3D models by comparing digital measurements to those obtained from the skeletal elements and to streamline the application of curvature analysis for the estimation of assigned sex from the GSN. This study utilizes the left and right os coxae from 15 skeletal individuals (5 females, 10 males) from the Boston University Chobanian & Avedisian School of Medicine. A Fujifilm X-Pro2 and Fujifilm 35 mm prime lens captured 123 images per element, which were processed in Meshroom by AliceVision® to create a 3D textured mesh. The mesh was exported into Blender for cleanup, scaling, measurement, and curvature analysis. The measurements were between 96.54% and 99.94% consistent across methods and observations. The consistency between digital metric observations increased by an average of 0.07% when compared to the consistency of the dry bone measurements. Additionally, curvature analysis of the GSN correctly estimated the assigned sex of all os coxae in the sample. This study demonstrates that photogrammetry is an accurate and reliable method for the digitization of remains that enables analytical techniques to better capture skeletal variation compared to traditional methods.

No description available.

Macro Planning of construction projects is among the most challenging tasks faced by the project team. Decisions made during this process have a tremendous impact on the successful execution of the project from its early conceptual phases, through the project construction and completion. For a large majority of construction projects, the current planning practices remain manually based. General and project specific data are communicated among project participants through design drawings in a 2D paper-based format. Due to the interdependence between the different elements and the large amount of information that needs to be manually processed, the current manual implementation approach is very difficult to undertake, and imposes a heavy burden on the project team to carry out the planning process. Various research efforts have been undertaken in an attempt to capture current planning techniques and allow for the development of new innovative and automated ways in planning. The developed planning systems are characterized as responsive decision systems, relying mainly on programmed knowledge and heuristics for decision making, hence reducing or eliminating the role of the human planner. This research presents the framework for a new interactive planning environment called the Virtual Construction Environment (VCE) that supports the thinking process of the project team during the macro planning phase of design-build projects. Unlike previous responsive-type systems developed, the approach utilized in the VCE is supportive to the project team enabling them to be an active participant in the decision making process. The main purpose of the VCE is to assist the project team during decision making, by providing pertinent information necessary for making appropriate decisions in a structured format. This information may be organized, stored, and retrieved by the project team whenever needed during the virtual sessions. The VCE also provides the project team with appropriate tools to test different work execution and site layout planning scenarios early during project development. During the virtual sessions, the project team reconstructs the facility by bringing graphical elements together. The project team's movements and interactions are recorded to capture their thinking process on how to construct the facility (i.e. sequence of major assemblies). Other project participants can retrieve recorded decisions for further review or modification. The project team is also able to specify construction methods, and allocate resources required for the implementation of major assemblies. The VCE guides the project team to perform these interdependent planning functions interactively and concurrently. Using system graphical libraries, major equipment and temporary facilities can be superimposed and displayed as graphical objects for site layout planning. This enables the project team to visually check for space and accessibility conflicts during different virtual construction time intervals. In order to define required information in the VCE, the author has developed a MAcro Planning Information Classification (MAPIC) model under which information required for macro planning decision making could be classified and organized in a structured standardized format. The project team may then retrieve and utilize this information whenever needed during the virtual sessions. A prototype computer tool is developed to illustrate the framework of the VCE. The computer prototype is implemented using available commercial software tools. / Ph. D.