X-ray CT scanning is effective to inspect the geometry and positional relationship of parts in assembled objects. For further application of CT data, for example for measurement or design, parts segmentation and extraction is necessary. However, it is still difficult to correctly separate two parts sharing a narrow gap on the CT data. This issue can be reduced by a use of the design data (CAD mesh) of the object. In order to automatically recognize the parts in the CT volume using CAD data, precise alignment of the two data sets is necessary. However, the shape of real products may be different from the design because of a modification during manufacturing or deformation by assemble. This shape deviation between the two data makes an automated alignment difficult. With this reason, currently, the alignment of CAD data and CT scanned data is conducted depending on the human eye, and it increases time and human costs, and the results highly depend on the operator. In this study, we propose an algorithm to align a CAD mesh and a CT volume with a higher precision than an existing algorithm. The proposed algorithm first extracts a surface mesh from the CT volume, and then move the CAD mesh so that it matches the extracted mesh by the transformation computed by an optimization regarding the vertex positions and the orientations of the normal vectors of the CAD mesh and the gradient vectors of the CT values. Experiments show that the proposed algorithm achieved the alignment accuracy in average almost same as the voxel size.

In human communication, nonverbal behaviors such as facial expression, body movement, eye-gaze, and pupil response play an important role in realizing activated interaction and communication. In our previous study, an activated interaction through avatars was analyzed, and an estimation model for the activated interaction was developed by introducing the concept of heat-transfer engineering. This model estimates the amount of heat in the communication field as the degree of the activated interaction based on an assumption that each talker's heat is transmitted to the communication field. However, it is hard to estimate interactions including the conveyance of eye-gaze because the model has no parameter of the eye-gaze. It is important to share the conveyance of mutual eye-gaze for producing the activated interaction in voice communication. Therefore, it is expected to develop a model that estimates the conveyance of eye-gaze. In this paper, focusing on eye-gaze which plays an important role in producing the activated interaction, we developed a model that estimates the degree of conveyance by the talker’s eye-gaze in voice communication. This model estimates the degree of the interaction-activated communication as heat transfer by combining the speech input and the eye-gaze input.

Although the number of traffic accidents has been decreasing in recent years, the number of automobile accidents caused by pedal error has remained at a certain rate. However, the mechanism of the occurrence of such accidents has not been revealed. By using a driving simulator, it is possible to obtain driving behavior data related to pedal errors safely and efficiently, which may help to investigate the mechanism of these incidents. It is considered that an automobile driving simulator with head-mounted display can reproduce more realistic driving behaviors without any discomfort than a conventional simulator that presents images on a planar display. A head-mounted display with a head-tracking function enables the user to observe all directions of the field of view in response to the movement of the head, which results in a highly immersive image presentation. In this study, we analyzed the effects of changes in driving posture of look back on pedal operation, focusing on the accelerator and brake pedal operations.

During the COVID-19 pandemic, most of the interpersonal communication is conducted remotely. The insufficiency of discussion on achieving tele-immersion with others in a real place is also revealed. There is a need for improving the remote participation experience of communication in real places to meet the spiritual needs arising from the increase of remote communication in daily life.

We analyzed the elements which make up the unique value that could only be achieved through onsite communication, and the importance of spatial information is emphasized in this study. A concept of realizing a greater sense of freedom in spatial cognition and expression during the communication with others in real space is proposed by using avatar robots. We then designed a system that enables people to build stronger connections with the communication space in reality through remote connection. The prototyping, verification and validation of this system was conducted using Pepper robots and we discussed the directions for further improvement.

In recent years, the spread of guiding robots and household robots has increased, and research on communication between humans and robots has been actively conducted. In particular, there are several communication robots that look similar to humans and can interact with humans, so they are expected to be able to communicate like humans. However, at present, it is difficult to say that communication robots are able to communicate naturally with humans. In this paper, we discuss the factors that make people like robots, based on evaluation experiments of multilingual guidance using a small robot Sota and a humanoid robot Pepper, aiming at natural communication between people and robots.

In recent years, electric vehicles have become increasingly popular. To operate safely them requires the recognition of the appropriate state of charge (SOC). However, the different between the real SOC and the predicted SOC can lead to problems such as missing the timing of charging. In order to solve this problem, we modeled the relationship between the driver's predicted SOC and the perceived SOC by man-machine model. We also introduced the informationistic free energy as a measure of the likelihood of the perceived SOC and modeled the relationship with the emotions of security and unease. The difference between real SOC and predicted SOC is defined as prediction error. Two models led to a hypothesis about the relationship between prediction error and emotion. The experiment show that subjects felt more uneasy in the larger prediction error condition. The experiment confirmed that emotions are affected not only by prediction error but also by the variance of recognition.

Autonomous mobile robots are essential for us in recent years. Living with robots become more common, and we humans and robots are required to respect each other. For smooth interaction between humans and robots, human predictability for robot motions desired to be improved as if human can easily predict another human’s motions. We model human prediction for autonomous mobile robot motions, focusing on acceleration. We hypothesize that a human predicts the acceleration of a robot based on observed mass under the motor driving force as a constant latent variable. In the hypothesis, we propose inverse relationship between the acceleration and observed mass. In addition, we formalized surprise from unexpected motions based mathematical model that surprise increases when prediction error gets bigger, and prediction uncertainty decrease the effect of prediction error on surprise. We also investigated a bias toward perceived acceleration, which called expectation effect. We conducted an experiment using 3D animation with a mobile robot model. The result supports the prediction model and these hypotheses.

In modern society, recession and restrictions on going out have affected the quality of life; stress-related illnesses, such as depression and psychosomatic disorders, are increasing, and spending time alone in one's room has become a cause for concern. Candles have been used as sources of light for a long time; in recent years, a large proportion of women use candles as indirect lighting in public spaces as well as in their own rooms for relaxation. However, there is growing concern that candles may cause fire accidents; therefore, the use of electric candles has increased. In this study, the differences between the relaxing effects of electric candles with and without fluctuations are examined via fingertip plethysmography by assuming a windless space. Pulse waves and questionnaires were used for evaluations. We also examined whether similar relaxing effects could be achieved for gender differences. The results showed that fluctuating electric candles were expected to have relaxing effects. The responses to the impression evaluation questionnaire showed that there were no significant differences between the effects achieved with electric and wax candles.

This study proposes a design process to achieve the required comfort environment by effectively using sensory information acquired by five senses. This method can improve the thermal environment and save energy at the same time. According to previous studies, sensory information influences people's perception of warm and cold. However, the evaluation of the influence in a complex environment consisting of multiple sensory information has been insufficient. In contrast, this study evaluates the combined influence of visual and auditory information on the sense of warm and cold by measuring the change in perceivable temperature difference when subjects are given sensory information. In the measurement experiment using subjects, the perceived temperature changed when images of a snowstorm and a bonfire were given as sensory information. The electroencephalography was measured at the same time. The results showed the possibility of quantitatively evaluating the complex environment by biometric measurement.

The total annual radiation on photovoltaic arrays is affected by the position of the sun and placement angles of a photovoltaic array that are included tilted angle and azimuth angle. In addition, it is also affected by the shape of the site and the position of photovoltaic arrays. The photovoltaic arrays generate electricity by aligned placement at the large-scale photovoltaic generation plant. On the other hand, in the case of the small-scale and complex shape of the site, there is a possibility that the aligned placement of photovoltaic arrays causes blank spaces in there. Therefore, these sites are expected to increase the total annual radiation on photovoltaic arrays according to the arbitrary placement of photovoltaic arrays. This study examines the placement optimization of each photovoltaic array at the site of small-scale and complex shapes. The placement parameters are used tilted angle, azimuth angle, and placement position of each photovoltaic array. The photovoltaic arrays with different placement parameters are optimized simultaneously by the genetic algorithm to maximize the total annual radiation of photovoltaic arrays. As a result, the photovoltaic arrays are arranged at sites of arbitrary shape, and the total yearly radiation of photovoltaic arrays was maximized.

In this paper, we apply a short-term prediction method using an RBF network to conventional feedback control. The conventional short-term prediction method is fixed on the absolute coordinate system and the prediction accuracy is insufficient. In this paper, in order to improve the calculation accuracy without increasing the amount of calculation, the coordinate transformation by affine transformation is introduced. Therefore, the training data on the absolute coordinate system is transformed into the rotating coordinate system using the affine transformation, and short-term prediction using the RBF network is performed on this rotating coordinate system. As a result of numerical calculation, it was shown that the method incorporating the affine transformation improves the prediction accuracy. Comparing the conventional feedback control and the proposed method with the spring mass model of the one-degree-of-freedom system, it was found that the proposed method converges faster.

As for global warming, it is currently in progress and countermeasures are being taken to cope with it. As for forest resources, the population has increased explosively with the industrial revolution, and as a result, forests are being cut down for use as building materials and to expand farmland and residential areas. In addition, deforestation has made it impossible for indigenous species to survive, resulting in habitat displacement and, in severe cases, extinction. In order to solve the problems of soil and ecological preservation caused by bamboo forests, we believe that we can solve environmental problems from the perspective of reusing waste wood. Bamboo is a material with excellent elasticity and is suitable for bending. In this research, we aim to create an environmentally friendly and affordable structure using waste wood that is left after being cut down, and analyze it using 3D CAD software such as SolidWorks and Fusion360. We will design a plastic greenhouse structure using bamboo material that can withstand snow and wind. The goal is to be able to create a model and design a structure systematically by applying the design of experiments as an element. In addition, we will compare the design using the design of experiments method and the structure using topology optimization, and optimize the design to expand the range of utilization of bamboo materials and to further improve the plastic greenhouse structure.

The purpose of this study is to increase the number of reproductions of point clouds when reconstructing a branch-shaped complex structure from moving images. It is assumed that the camera mounted on the small UAV will be used for shooting. For deciduous trees, the structure of trunks and branches is reconstructed from images. VisualSfM is used for the 3D reconstruction. We will try to improve the number of reproductions of 3D reconstruction by increasing the amount of point cloud data by improving the image quality of captured moving images and adding moving images taken from different angles of view.

Currently, the development of autonomous driving technology is being actively carried out in the world. There are already passenger cars that perform Level 2 autonomous driving, and Level 3 passenger cars are just around the corner. In this way, there is a strong tendency to aim for driverless driving worldwide, but if such automatic driving becomes commonplace, the enjoyment of driving a driver will be lost. In this research, we will create a driver model using machine learning and develop a model learning method so that the driver can drive with peace of mind and enjoyment. The ultimate goal is to create a system that allows people who are not good at driving to enjoy and safely drive by using a simulator to think of the driver model as an actual person.

There is to develop software to support the design of running shoes，one of the most familiar sporting goods．Currently，we are focusing on the upper part of the design．The upper part is made of a knitted material，which is treated as an anisotropic Hyper-elastic material in the finite element analysis．The parameters of the Hyper-elastic model are identified from experimental data．Since manual identification is time-consuming，we will create software that implements automation．Several optimization methods that could be used in the identification of the parameters were investigated.

Dynamic FEM analysis is valid for designing rotating machinery to reduce its vibration problem when we may ensure enough accuracy of the analysis. Surrogate multiple objective optimization method is one of the most effective methods for structural identification improving the FEM analysis model of a structure to adjust the natural frequency analysis results to the experimental results. In this study the structural identification method is applied to coil mounted stator core of an induction motor to determine the Young’s modulus of the principal components on the FE model minimizing the analysis errors of the natural frequencies of the 2-lobe and 3-lobe circular modes to the corresponding experimental results. The accurate FE model of the end-windings is obtained by this method.

This paper deals with uncertainty of moments of multivariate normal distribution that design variables in reliability-based design optimization follow. Specifically, we assume that the expected value vector and the variance-covariance matrix belong to given closed convex sets, and require that structural reliability in the worst case is no smaller than a target value. It is shown that this performance requirement can be reduced to some deterministic constraints. A simple numerical example demonstrates a potential that the presented formulation can be handled within the conventional framework of deterministic optimization.

In the multi-objective simultaneous satisfaction design method (set-based design method) based on the preference degree, the distribution of preference numbers regarding the performance as the ability of the function and the design variable as the influential factor have conventionally been based on empirical knowledge. In addition to this, in this study, I investigated a method based on directed graph theory (adjacency matrix and governing matrix) and a method based on the frequency of actual product usage as characteristics of the design variables. In particularly, for the of cases of graph theory and the frequency, the methods expressing the preference distribution are shown using actual concrete examples, that are driving conditions of a car and material properties, respectively.

A concept space is a graphical tool to map keywords and images of design visually onto a metric space. While it stimulates designers’ability of problem finding and ideation, there are open issues toward design concept generation. This paper proposes a framework of design concept generation with an extensional concept space (ECS). ECS is a metric space of the characteristics, i.e., function and physical attribute, each of which is a set of entities that fall under its definition. The concept distant measurement method based on word2vec with Wikipedia corpus is adopted to extract concepts from documents of design cases and to build a concept space. The framework facilitates a designer to explore a concept space that contains the text and image information with iterative operations, such as function development, and create a new design. This paper demonstrates a case study of chair design to verify the effectiveness of the proposed framework.

In this study, we investigated a judgment method that classifies the prediction of driving behavior as correct or incorrect using a histogram of the prediction probability of driving behavior. XGBoost was used to judge the correctness of the prediction of driving behavior. The relative frequency of the histogram and the prediction class of driving behavior were used as the input features to XGBoost. The evaluation results of the judgment method for predicting driving behavior showed that the true negative rate was 63 %. This suggests that the proposed method is effective when incorrect predictions need to be excluded as much as possible.

In the manufacturing industry, the transfer of knowledge of skilled persons is an issue because the number of skilled persons is decreasing. In this study, as a basic study for constructing a system for transferring skilled person knowledge using gaze measurement, we tried to extract the knowledge of the skilled person by comparing the gaze of skilled person and unskilled person. In this report, we compared gaze of participants for examining the phenomena from fluid simulation results. Participants were classified as expert and beginner. As a result of gaze analysis, experts spend time observing the areas of each vortex. And it is cleared that experts spend a lot of time observing ellipsed vortex which is difficult to find for participants.

In recent years, application of virtual reality (VR) for work-related training at production sites has progressed rapidly, but there is a problem of insufficient training owing to the lack of educators. The aim of this study is to verify automatic motion analysis using a VR-based training system such that trainees can recognize and learn critical information for self-improvement at work. In this study, we used video cameras to capture an operator’s motions and actions during the outer rounding and end-face cutting operations on a lathe. For the motion analysis, we used a learning model for time-series segmentation, which extracts features from video images and predicts the class of each frame. We prepared six labels, including four labels for advanced and penetration-type movements as well as two labels for static and transporting empty. A model was created and trained with 90 video data samples of several minutes each, and the training results showed more than 80% success with several metrics, such as accuracy, edit distance, and segmentation F1 score with an overlapping threshold of 50%. These results suggest the feasibility of application of the proposed scheme to training systems.

In recent years, there has been a need for shorter analysis techniques in order to reduce the time required for design optimization. For this purpose, approximation by polynomial regression, which is a mathematical representation of the analytical structure, can be considered, but the difference from the real structure is large and the error is large. In addition, an approximator based on machine learning requires a lot of time to create training data, so a method that can be evaluated accurately even with a small amount of training data is essential. Therefore, by combining a neural network approximator that mimics polynomial regression and a neural network trained on errors with unknown governing equations, we have shown that the accuracy can be improved even with a small amount of training data.

The machine learning approximator of the simulation output has a large error when the training data is small because the training data is discrete. Therefore, we thought that the accuracy could be improved by additional learning based on the concept of active learning. In response to this, we added learning points to the middle points of the neighboring tents of the learning points with large errors, but so far we have only been able to derive points that improve the accuracy by about 70%. In this study, we found that by recognizing the shape of the learned approximator as a distribution of curvature, we can propose an appropriate additional training point with a probability of 90%, which is expected to improve the accuracy.

Recently in the field of design, it is becoming more important to generate innovative ideas of "what to create" at early design stage than ideas of "how well to make". Ideation informatics to enhance idea generation by adding systematicity and exhaustiveness of information technology to human intuition should be a prospective approach to the problem. In this research, a ten sentence pattern model is proposed as a computable description of function and user experience (UX) at early design stage as a fundamental technology for ideation informatics by extending the English basic five sentence patterns. An XML format is designed to add semantic information to words, phrases, and sentences described in the ten sentence pattern model by using concept dictionary, i.e., concept identifiers of EDR electronic dictionary and synsets of Japanese WordNet, and software to calculate their semantic similarity is implemented in Python. Then, based on the knowledge of cognitive neuroscience that "human creation does not create something out of nothing, but the memory of the past is the basis of creation", a database of functions and UXs of existing products and services is prepared as an extension of human memory. Using the database, experiments to coming up ideas for solving problems related to the new coronavirus. From the obtained results, the effectiveness and possibility of the proposed method are confirmed.

Huge number combinations for distance among parts is required to obtain the clearance between the parts by calculating the minimum distance of surfaces on the parts. Various polygon-based high-speed solution methods have been proposed, but they have accuracy problems. In this paper, we propose a new method for calculating the distance between two surfaces that combines the approximate solution method using voxels and the high-precision solution method using NURBS surfaces to satisfy both accuracy and speed. The method was implemented in the "Design Insights CAD system", which is an automatic verification tool for design rules. This method was applied to three verification models, and was found to be about 10 times faster than the conventional method using only a high-precision solution, while keeping the accuracy.

We are development “Insight CAD system” as one of the design support technologies.The challenges in 3DCAD design are following two items, 1) To realize visully checking of design rules to reduce checking time of which CAD model complexity caused by increasing number of parts. 2) To prevent of rework caused by impossible assembling in manufacturing stage. Therefore, the purpose of Insight CAD system development is to check design rules automatically, and to give the information about the violation part and the error list for designers. We have been developing the technology to digitize the design rule efficiently for the purpose of deploying this system to various products. In this paper, we report the technology which automatically estimates the parameters in the digitization rule that are different for each product.

We developed a part characteristic design technology for a metal additive manufacturing using topology optimization. In the part design, it is necessary to determine the shape in consideration of manufacturing constraints in order to obtain the required part characteristics. Although it is difficult to obtain the appropriate manufacturing conditions, we have been able to predict deformation, stress and reduce dimensional errors in the heat shrinkage deformation during melting and solidification in the additive manufacturing shape design technology we have worked on. On the other hand, manufacturing constraints are related to factors other than heat shrinkage deformation. If there is no knowledge such as past results, we will experiment with elements as necessary to investigate manufacturing constraints. The requirement of the verification model of the technology is that the eigenvalue exists within ± 10% of the required frequency when the material is stainless steel and the outer surface is fixtured. We have established a process to design component characteristics by combining topology optimization considering the possible manufacturing area of the metal additive manufacturing, and the evaluation of manufacturing feasibility using a metal additive manufacturing process simulation, reflecting the experimental results of the elements. We confirmed that we can apply the development technology to the verification model and obtain a structure that can be manufactured and have the eigenvalue of the aim. In the manufacturing of the spring, it was confirmed that the manufacturing or not is different depending on the layout.

We are proposing the concept of ‘Digital Triplet’ (D3) to support manufacturing system engineers in executing engineering processes, including Kaizen, by utilizing Digital Twin. The goal of our research is to verify the feasibility and advantage of D3 by embodying D3. For this purpose, we develop a learning factory based on the D3 concept. First, this paper describes the requirements of the learning factory. Next, we illustrate how to implement the learning factory and the results of the implementation. The developed learning factory consists of the three parts that correspond to the three worlds on D3, i.e., the physical, cyber, and intelligent activity worlds. Finally, this paper describes the difficulty we faced in the development and the insight gained from the difficulty.

X-ray CT, which is widely used as a non-destructive inspection method, has the problem of long measurement time. If the measurement time is too short, the transmitted image will be blurred or noisy, and the quality of CT volume will be reduced. Therefore, there is a trade-off between time reduction and quality. In this research, we aim to develop a method that can both shorten time and improve the quality. We train CNNs to improve image quality on a previously obtained dataset, and then apply the CNNs to another dataset. With the loss function proposed in this study, we can achieve high quality output results of CNN and we evaluated it with quantitative metrics and visuals.

Industrial X-ray CT is nowadays an essential tool to inspect industrial assembly without disassembling. However, these CT volumes are usually composed of a huge number of voxels without any information representing its parts, which prohibits visualizing only the parts that the users are interested in. On the other hand, it is considered effective for this problem to leverage geometric features extracted from the CT volume. According to this thought, this paper introduces a new geometric attribute, which we refer to as a degree of local geometry (DLS), involved with each voxel and investigates on its effect for visualizing CT volume. The DLS represents a dimensionality of the manifold around a voxel, e.g., that for a surface will be 2. We experiment its effect in visualizing CT data by installing the DLS on voxels into a sparse voxel octree (SVO), which significantly reduces data size and computational complexity. we demonstrate that the DLS works significantly well to selectively visualize small parts inside several test assemblies, such as a set of primitive shapes and a radio-controlled car.

We propose an approach to generate geodesic grids for approximating freeform surfaces. We employ the accurate numerical computation of high-order derivatives along geodesic curves on surfaces to determine the width of the geodesic strips. It computes derivatives of arbitrary order from the result of the numerical method employed for computing the geodesic. We demonstrated the effectiveness of our proposed method by applying it to an apple model.

We show that the surface of an umbrella has negative Gaussian curvature, and due to its symmetry, the surface of the umbrella between bones can be modeled as a piecewise bilinear surface. We investigate various differential geometric properties of the umbrella surface and demonstrate the effectiveness of our proposed geometric modeling of umbrella surfaces.

Mostly, umbrellas are designed empirically. We consider using numerical analysis for an umbrella design. We use T-spline to represent the model to have smooth basis functions as a mesh for structural mechanics. The challenge is to represent the ribs as a one-dimensional model on a cloth represented by a smooth basis function. We use the T-spline for this challenge. In this research, this is realized by applying the T-spline.

The final goal of proposing and organizing this session is to develop a new economy, where everybody will be happy and enjoy life. Since the Industrial Revolution, we started to work for others. Until then, we worked for ourselves. In the old days, we started engineering to make our dreams come true. We worked alone for ourselves in pursuit of our dreams, which vary from person to person. And then we started to barter things. But still we worked for ourselves. We wanted to satisfy our own material needs. But the Industrial Revolution changed the whole scene. It introduced division of labor and we started to work of others, i.e., for external rewards. But human needs shifted from material to mental satisfaction. We would like to actualize ourselves. And the maximum satisfaction and feeling of achievement cannot be secured unless the job is internally motivated and self-determined. The Industrial Society is getting close to the ceiling and many issues are emerging. Now is the time to develop a new society for the next generation. In addition to these issues of the Industrial Society, the current economy is monetary. So, any time there are winners and losers. If we can develop a self-satisfying society, where people can self-sustain and can enjoy life each in his or her own way, then we can satisfy the self-actualization needs and make everybody happy. To achieve this goal, we need to create a new framework for developing enjoyable engineering.

Movement is essential for living things. That is why living things are called creatures. They create movement to survive. But the real world is changing rapidly. Yesterday, changes were smooth, so that we could differentiate them and predict the future. But today changes are sharp, so we cannot predict the future anymore. And our world was closed with boundary yesterday. But today, boundaries disappear, and it is open world. So, it is difficult to control our world mathematically. And what make the problem increasingly difficult is materials are getting softer and softer due to the rapid progress of material engineering. Therefore, instead of control, we need to coordinate many elements to adapt to the rapidly and extensively, and unpredictably changing real world. In short we need to coordinate our body movement to cope with the current environments and situations. This needs a wide variety of information and strategic decision making is called for. But if we remember that we used to learn how to cope the real world by directly interaction, when we were babies. And it should be emphasized babies enjoy this interaction and amplify their capabilities of perception and movement. Regrettably, we have not paid too much attention to instinct. But babies perceive and move, based on their instinct. And they enjoy their life. So, we should pay more attention to our instinct and make our life more enjoyable.

The Industrial Society since the Industrial Revolution is getting close to its ceiling and many issues are emerging. Now is the time to prepare for the next generation society. This paper discusses what engineering we should develop for tomorrow. In the current industrial society, we have been working for others and engineering has been full of sophisticated technologies. But this framework consumes tremendous amount of energy and we are not truly happy, because we work for others. We feel most satisfied and are filled with the highest sense of achievement, when we do a job which is internally motivated and which we self-decide. Thus, the self-sustaining society is desired for the next generation. To achieve this goal, engineering should move away from working for others to enjoying our own life. We should develop a pleasurable world. This is nothing other than establishing true service. Service in the true sense means making every stakeholder or everybody concerned happy. To achieve this, we need to coordinate many elements and let them work together as a team, as IoT proposes Things Team, where man and machine work together on the same team. In fact, orchestration is now getting attention in computing. Therefore, we are now ready to move ahead toward developing an enjoyable life. And it should be emphasized instinct, which has been ignored up to now, will play an important role in the next generation.

This paper clarifies evolution of services and innovation drivers for service creation. Then, it identifies the skills required for future creators of digital native services and discusses educational programs for them at universities. The curriculum includes essential capabilities; how to develop mindsets of design thinking and systems thinking, how to learn design methods of system of systems and systems engineering processes, and how to form a team to carry out a service development project. The findings through the programs reveal that conventional systemic approaches need to incorporate emergent ICT environments i.e., analytic approaches of data science and artificial intelligence.

Crop diseases are one of the factors behind the decrease in crop sales. Since the kind of crop diseases is large and there are diseases that damage other crops, early detection is required. Not only detecting disease strains needs much, but it is also difficult for new farmers to distinguish them; it is necessary to automate the early detection of diseased strains. This paper proposes a disease strain detection method using CAE (convolutional autoencoder) for onion strains. Since CAE learns only from strain images that are not disease, it is effective when there are few disease strain images. In the computer experiments, two types of image cropping methods are compared; the first type is to cut out manually the disease strain so that it appears in the center of the image. The second type is a method of automatically dividing images. As results of computer experiments, the effectiveness of the proposed method is confirmed; the discrimination rate is higher in the first cutting method than the second one.

In recent years, the number of elderly people living at home (requiring nursing care or living alone) has been increasing due to the aging of society. As the elderly stay at home for long periods of time, their connection with the local community becomes less and less, and loneliness becomes a serious problem. In our research group, we are developing a system that allows elderly people to interact with the virtual agent. The purpose of this paper is to develop a method for generating personalized dialogue in order to realize continuous and empathetic dialogue. As an approach, morphological analysis and sentiment analysis are performed on the dialogue content (text) obtained from speech recognition. Then, we present a new method for generating personalized dialogues based on related conversation log summarization. In this way, it is promising to realize smarter personalized dialogues that do not rely on manual creation.