ABSTRACTS of TECHNICAL PRESENTATIONS

Japanese industrial products have been competitive in the global market due to their high quality characteristics such as fault-tolerance, energy-efficiency and assured safety. These factors are still fundamentally important, but they hardly alone attract customer interest because products with same level of such qualities are brought to the marketplace from many companies in various countries in the last decade.

Under such global competition, it is critical to the industry in our country to intensify another category of product quality, that is, delight quality which invokes customer’s psychological feelings, such as those exciting, comfortable and happy. In Japanese, the word ‘Kansei’ is often used to refer to the delight quality.

We started a five year project funded by the government last year to develop a design environment to support the delight design. We recognize no such environment has been built and used in the industry though there are pieces of fundamental design technologies related to the delight design. We rather aim at developing a platform by integrating these technologies to cover design stages from the conceptual design to the detailed design. This platform is a package of design methodologies, software design tools and design guide lines. In my presentation, its key concepts and basic technologies of the platform will be introduced.

A methodology, which allows an explicit inclusion of crew risk and safety considerations into the preliminary design of small ships have been developed. The methodology combines the traditional preliminary design spiral of ships with a method to include crew safety considerations in a knowledge based platform, called Echidna.

A ship, from the onboard crew’s point of view, can be thought as a work and a living platform. It has been reported that motions of this platform is the most prominent contributory factor on board safety and performance of the crew. Hence, two different methods have been developed to reduce the impact of platform motions. As a case study, the methodology is applied to the design of SWATH ships. The study presents a mapping of crew safety to the sea state and vessel parameters, such as size and cost.

Some measurements of surface waves for the direct measurement of ship wave resistance, with smart cameras are also to be discussed.

 There are image processing methods that minimize the total variation in the image. These methods assume that the images are typically composed of low frequency components, and the structure of the objects is typically ignored. In this work, we modify the classical total variation minimization technique that can process images with dominant direction. For this purpose, directional total variation is introduced where the sensitivity against variations in the image can be adjusted at a specific direction. Finally, applications of directional total variation are demonstrated for various image processing problems. The results indicate that directional total variation performs better than regular total variation minimization method given that the dominant direction in the image is well estimated. In addition, if the direction and the sensitivity are carefully adjusted, the performance of directional total variation increases.

Industry 4.0 is a program of German government and industry. It means the fourth wave of industrial revolution where smart technology including smart phones will play a role. Key components of the Industry 4.0 are smart factory, CPS (cyber physical system), and IoT (internet of things). Similarly in USA or other industrial countries including Korea the smart factory or the smart manufacturing is also emerging mainly based on the initiative from each government. Industry 4.0 is to make each device of a manufacturing system to be controlled through the internet. Sensors and actuators which are attached to the device can be accessed through IoT technology. To make many different and physically dispersed devices communicate each other we need to use the international standards. In this talk the background story of why governments are leading such industrial technology programs, the contents of the Industry 4.0 program including international standards which will be used in will be introduced.

Bioprinting for tissue or organ engineering is a relatively new technology where living cells with or without biomaterials are printed layer-by-layer in order to create three-dimensional (3D) living structures. In this presentation, biomodeling for three-dimensional bioprinting will be presented. The presentation will start with modeling for scaffold-based approaches. Scaffold modeling requires optimizing porous micro-structure of scaffolds for better tissue adhesion and proliferation. We have develop novel methodologies to model multi-functional tissue scaffolds by controlling and optimizing not only the porosity of scaffolds but also controlling the active biomolecules to improve tissue regeneration. In the second part of the talk, we will present biomimetic modeling of tissue constructs for scaffold-free tissue engineering. Here, biomimetic modeling to 3D bioprint live cells will be presented for printing totally biological tissue/organs.

Estimating the curvature tensor on the surface of a 3D-scanned object is fundamental and important for analyzing the 3D shape of a real-world object. Conventionally, the curvature tensor of a CT scanned object are estimated by using a reconstructed CT volume or a polygonized isosurface that are converted from the original scanning data called sinogram, i.e. the sequence of X-ray projection images. In this presentation, we propose a novel method in which the sinogram directly contributes for accurate estimation of the curvature tensor. Compared with the results obtained by the conventional methods, the discretization artifacts that is caused by grid-based resampling can be avoided by applying the proposed method.

The main task of product development is to develop a good product at lower cost and to bring it to market in a shorter period. Conventional computer-aided design and computer-aided engineering (CAD/CAE) systems are well established in this regard. However, although upstream design is particularly important in product development to add value and incorporate the required functions, it is difficult to apply conventional shape-based CAD/CAE systems to the upstream design stage due to the lack of design information at that stage. As a solution to this issue, we are proposing the development of a design framework called “1DCAE,” which can be applied to the early design stage of product development including the conceptual and functional design phases.

1DCAE is defined as the design concepts, methods, and tools that cover the early design stage. The output of 1DCAE is input as the design requirements to a conventional CAD/CAE system, which transforms the design requirements into an actual product image. The CAD/CAE results are then fed back to 1DCAE to verify the overall functions. The 1DCAE concept is expected to realize design innovation and more innovative products.

The main goals of Computational Anatomy (CA) are twofold: (i) the computerized identification of neuroanatomical changes in magnetic resonance images (MRI) that result from a disease and (ii) distinguishing the images of diseased subjects from those of the normal subjects (that have normal variation in the population). Both problems are difficult in the sense that making detailed measurements describing the brain morphometry leads to very high dimensional data. In this study we discuss segmentation, registration and statistical shape analysis methods for the problem of discrimination between structural changes in the brain due to a disease, say dementia. As illustrative examples, we discuss two recently developed algorithms: (i) Labeled Cortical Distance Map (LCDM) which is a powerful tool to quantize differences in shapes derived from MRI.  We consider some aspects of LCDM distances in relation to morphometry.  (ii) The Large Deformation Diffeomorphic Metric Mapping algorithm is a tool used to obtain the metric distances that quantize the morphometry of binary images of a tissue with respect to a template and a vector field which indicates the change of the original image towards the template. We highlight some of the challenges in the analysis of the very high dimensional data obtained from these algorithms and possible solutions.

Existing packing algorithms (such as strip or bin packing) take rectangular items as input and pack them into rectangular bins. In this research, the items that have to be packed are nonrectangular quads and can have larger size than that of bins, and the shape of the bins are polygonal. From polygonal bins, several quad regions ready for packing are produced which are called as candidate regions in this work. As the packed item can be larger than the candidate region, its portion outside the bin can have five different shapes: triangle, quad, pentagon, hexagon, heptagonal, octagon. The proposed split algorithm decomposes these shapes into quads each of which will be treated as item separately during the packing step. Packing of items into bins is done greedily based on a cost function which favors minimum waste of the bin after packing and minimum number of cuts formed the split algorithm. As a case study, a methodology for packing of a ship hull into input bins will also be presented in the talk.

Computed tomography is an inspection tool attracting increasing attention in industry for the last decade because of its ability to inspect an object with a non-destructive manner. In general the boundary is defined as an isosurface of a CT image, however, a boundary between two materials with different densities tends to incorrectly be located in the object with lower density. In this presentation we propose a novel algorithm extracting such a boundary accurately. In order to suppress such miss location of boundary, we adopt another definition of boundary, the maxima of the directional derivative of the gradient of CT value, instead of an isosurface of the CT image. We also take a sonogram as the input instead of tomogram for improving the accuracy of derivatives and avoiding the voxel artifact. Besides we adopt an unstructured mesh as an underlying structure on which a CT image is reconstructed, since it has strong advantages over a grid about the well representation of surface and less number of elements. We will also show the effectiveness of the proposed algorithm with some experimental results.

In the mobile machines such as robots or vehicles, it is unavoidable that to have a detection system that give the ability of detection of objects around and prevent the crash of the machine. The main principal in getting depth information from 2D views is firstly deceting the edges. By detecting edges it is possible to process very less data and reduce unnecessary information without deforming the structural properties of the image. Therefore it is one of the most important steps to detect objects and get depth information. The distance between the same pixels on different images is detected. Then, by using the stereo vision geometry, the deth information is calculated. In this study, C++ programming language is used to process the images.

Maxillofacial surgery is an operation to treat variety of diseases and defects occurred in the head, face, jaws and the hard and soft tissues of the maxillofacial region. Due to the specific anthropometry of the patients, the instruments and implants, which are used in this surgery and have standard shapes and geometry, can not be appropriate to be employed. Therefore, some customized instruments and implants are needed to use for the maxillofacial surgery. In most cases, due to the complex and nonstandard forms of these specific instruments, conventional machining techniques are not convenient for fabrication of such devices and some smart design and production techniques are preferred instead. In this study, a review of the design and manufacturing processes of the some patient-specific instruments and implants used in maxillofacial surgery is given. Also, as a case study, design, fabrication and clinical application phases of a novel computer-assisted, patient-specific sagittal split osteotomy (SSO) guide and soft tissue retractor are presented. Mandibular SSO is an operation performed for correction of the malpositions of the jaw bone due to dentofacial deformities. In this operation, sharp rotary tools are needed to be used for the osteotomy of the mandible and this can induce some complications. To decrease the occurrence of possible complications, it was designed and manufactured a novel patient-specific SSO guide and soft tissue retractor. This apparatus was first tested on a cadaveric bone in vitro. Then, a surgical application of the instrument was performed. This paper also discusses the first results obtained from the in vivo application of the apparatus.   

Standardization is crucial in design and facilitates the communication between customers and designers. Customers generally have difficulty to tell what they want exactly to designers because of lack of standardization. In this study, some high level adjectives are defined for identifying superyachts to overcome the communication problems using machine learning techniques such as SVM, ANN, K-means clustering etc. First, simple and abstract geometrical parameters are introduced which can also be used to design superyacht hull. Next, relationships between adjectives and parameters will be figured out using some statistical approaches. At the end, the determined high level adjectives will be converted into defined parameters. With the outcomes of this work, we expect that customers can tell what they prefer in their superyachts using the defined adjectives, and designers will understand customer preference and parameter values will be arranged to obtain a model expressed by customers. By using the parameters and adjectives defined in this study, high level CAD tools can be developed and even customers can easily generate variety of new designs by only modifying the parameters of existing designs.

Recently, demand of as-built plant 3D model is increasing due to its effectiveness in a revamping or dismantling project and 3D simulation of operation maintenance. However, the acquisition of as-built plant 3D model requires the reconstruction process of a 3D model from the laser scanned data of the existing plant. Although there are many commercial tools with user-friendly interface, the task of identifying and modeling each equipment with different size and type from the point cloud still requires a large amount of user input due to a complexity of the laser scanned data. There have been many studies focused on the automatic reconstruction of pipeline model which takes a large proportion of plant 3D model. In contrast, the studies on automatic reconstruction of equipment model has not been handled as much as pipeline model because of its relatively complex and inconsistent geometry. This research is focused on the reconstruction of equipment models using CAD database together with a 3D object recognition technique. The recognition of equipment highly depends on the segmentation of corresponding points set from the scan data. This research tackles the case of poor segmentation result and manages this issue through the proposed system which could successfully recognize the equipment from the stated condition.

The present study is being carried out to develop a high-level CAD tool for variant and fast generation of yacht hull design. The design process is based on the different attributes like strong, aerodynamic, classic or modern etc. for yacht hulls. The design modification for each attribute is based on the parametric techniques with different constraints. The parameters are employed as shape modifiers that will be modified for respective attribute. Fairness of the hull design is achieved by the sectional area curve defined at forward and aft perpendicular. In this method, the hull shape is partitioned into three sections; Entrance, Middle Body and Run. Three set of different shape modifiers will modify the profiles and feature lines of each section accordingly to the attributes.