Currently, the shortage of skilled workers due to the declining birthrate and aging population is becoming increasingly severe. One example of this shortage is the welding industry, where the introduction of welding robots is advancing to maintain operations. However, small and medium-sized enterprises find it challenging to implement such technology due to cost constraints. Therefore, it is essential to properly pass down the skills and expertise of seasoned technicians to younger employees and nurture their development. The SECI model serves as a framework for skill transfer, but there is little research that supports the internalization process, where continuous practice leads to the acquisition of skills. In this study, we investigate the impact of motivation on skill transfer and propose a support method. Assuming that the possession of certifications correlates with the level of skill transfer, we conducted a correlation analysis between motivation and the number of certifications held. The results indicated a moderate effect size, revealing that fostering favorable motivation could potentially enhance skill transfer. Furthermore, we developed a skill tree system using gamification as a method to foster favorable motivation and confirmed its effectiveness, demonstrating its potential to support the formation of desirable motivation. Additionally, we found that this approach could be proposed as a method for talent management.

The integration of human-centered principles has become paramount to achieving sustainable productivity and worker well-being. This research explores the application of the micro-meso-macro model as a comprehensive framework for human-centered manufacturing. At the micro level, individual worker interactions with tools and processes are optimized through ergonomic design, personalized training, and system-dynamical mechanisms. The meso-level addresses team dynamics, workflow optimization, and the implementation of shared resources to enhance departmental efficiency and collaboration. At the macro level, organizational success and long-term well-being are aligned to foster an environment conducive to continuous improvement and innovation. Through case studies and practical examples, we demonstrate how this multi-layered approach can lead to significant improvements in both worker satisfaction and operational performance. The micro-meso-macro model not only addresses the immediate needs of individual workers but also integrates these needs into broader organizational strategies, creating a cohesive system that supports both human and technical aspects of manufacturing.

In recent years, with the increasing number of inbound tourists to Japan, the development of a multicultural society has been progressing. In this context, it has become a significant challenge in product design to consider a diverse range of users who have different cultural backgrounds and lifestyles. Therefore, this study aimed to establish a product design support methodology that considers diverse users by predicting accidents using a machine learning model and visualizing the results with R-Map. Initially, in constructing the machine learning model, features such as age, race, gender, behavior, and product information were extracted from the NEISS 2023 dataset to predict the types of injuries, the injured body parts, the frequency of accidents, and the severity of hazards. The model was trained using a random forest algorithm, performing multi-class classification. Subsequently, the predicted results were mapped onto the R-Map to visualize the frequency of accidents, the severity of hazards, and the number of incidents. Although the improvement of the model's accuracy was identified as a future challenge, the study demonstrated that this approach could be useful for comparing accident scenarios among different personas. By adopting this methodology, it becomes possible to consider the diversity of users in product design, thereby contributing to safer and more inclusive products.

The objective of this study is to support decision making for designers by visualizing various solution space features in the design space in the early design phase including uncertain design information. We propose a method to make the interpretation of design features easier for designers by evaluating the solution space extracted from a high-dimensional design space in terms of the size of the solution space, independence among design variables, and constraint achievement rate. By facilitating the design interpretation of the solution space, it is expected to allow the designer to select a solution space that reflects his or her intentions, thereby reducing rework in the later stages of the design process. The proposed method consists of five steps: derivation of the design solution space, extraction of various areas, evaluation of each area, displaying area characteristic pareto charts, and comparison of areas. This paper demonstrates the effectiveness of the proposed method as a decision support method by applying to a simulation model of an electric vehicle motor as a case study.

A qualitative comparison of the effort required to obtain design solutions using three representative simulation techniques that are essential in current product design technology is conducted. These techniques include the conventional CAE simulation method, the surrogate method combined with machine learning, and the set-based method. The comparison is mainly based on a qualitative assessment considering the characteristics of each method as a technique for representing the performance behavior of the product under design. For the set-based method, which is not yet widely adopted, an example problem involving the thermal conduction and stress design of an electronic substrate with two IC chips is used to facilitate its explanation. In conclusion, it can be stated that all methods are based on CAE simulations grounded in 3D-CAD modeling, and among them, the set-based design method requires significantly less design effort to obtain solutions. Therefore, in terms of design workload, the set-based method can be considered the most efficient．

Adaptive cruise control (ACC) is one of the critical elements of vehicle performance. To ensure the quality of ACC performance, comprehensive evaluations that control both complex test scenarios that reproduce market conditions and vehicle behavior is required. However, it is difficult to evaluate all combinations of test scenarios using real test vehicles within limited development resources. Furthermore, it is necessary to determine Electrical Control Unit (ECU) parameters while considering multiple performance trade-offs. This paper proposes a new automatic screening and exploration system for ACC, incorporating Bayesian active learning (BAL). This system consists of two automated elements: an automatic evaluation system and an automatic exploration system. In the automatic evaluation system, the behavior of ACC is automatically evaluated in real-time simulation using Real Car Simulation Bench (RC-S). In the automatic exploration system, the worst condition screening evaluation of ACC performance and the exploration of the feasible region of design space for ECU parameters using BAL are conducted. As a result, it becomes possible to make the evaluation process more efficient through closed-loop evaluation, thereby improving ACC performance. In this study, an example of data comparison between RC-S and a real vehicle driving on a test course is demonstrated to show the effectiveness of the proposed system.

Due to rapid changes in the market and various customer values, it is necessary to shorten the development period of automobiles. Vehicle performance has been improved mainly through experimental analysis and CAE (computer-aided engineering). In the increasingly rapid development of vehicles, machine learning is taking the lead. Vehicle performance prediction usually involves constructing surrogate models using design parameters and CAE results. However, three-dimensionally complex vehicle shapes cannot be fully represented by design parameters. Additionally, reducing vehicle drag has become even more important with the rise of battery electric vehicles. While balancing design and vehicle performances, complex three-dimensional shapes are explored to find optimal solutions, which requires a substantial amount of effort. For predicting vehicle drag performance, a method using Variational Autoencoder (VAE) is proposed to predict the shape of the vehicle front bumper side, drag coefficient, and flow field on the side and rear of the vehicle. With the proposed method, the drag coefficient was predicted with a maximum error of 0.012, an average error of 0.002, and an R2 value of 0.88, demonstrating good agreement with CFD. Additionally, the predicted velocity magnitude distribution on the side and rear of the vehicle is similler to CFD results. By creating a scatter plot (map) of the latent variables of the proposed method using principal component analysis results, it was found that the direction of increase in the first and second principal components corresponded with the increasing trend of the drag coefficient. Using this map, it become possible to predict the drag coefficient, velocity magnitude distribution on the side and rear of the vehicle, and vehicle shape derived by the proposed method. By using the proposed method, it is possible to suggest three-dimensional shapes that were not represented by traditional design parameters, making it easier to balance design and perfomances and thereby facilitating the search for optimal solutions.

Today, multi-objective optimization using simulation is widely used. Multi-objective sequential approximation optimization is used to perform multi-objective optimization problems with a small number of simulations. In the past, a large amount of simulation data was required because a uniform method of adding training data was used, despite the fact that the dominant effects were different. This study simplifies the problem by dividing Pareto solutions with different dominant effects. Pareto solutions with the same dominant effect are assumed to be close in distance in the design space, and the Pareto solution is divided by clustering to create a Pareto solution with a simple shape. We then describe how to evaluate and augment it with a small number of data. A multi-objective successive approximation optimization method using these methods is proposed, and the multi-objective optimization is demonstrated using a power device as an example. The results show that clustering by distance in the design space is equivalent to dividing the Pareto solution by the dominant effect.

Wire and cable sheathing materials are a mixture of PVC and EM (eco materials). When recycling plastics, it is important to separate different materials into the same material because mixing different materials decreases the quality of the recycled product. Hydrocyclone equipment can separate plastic particles with different densities using specific gravity water. In this study, a particle method model of the hydrocyclone system was developed to improve the separation efficiency of PVC and EM, and the results of estimating the recommended range of parameters that enable significant separation are reported.

The present paper deals with the vibration and damping characteristics of viscoelastic laminated sandwich composite structures, the composite skins and the viscoelastic core of which are manufactured separately. A multi-objective optimization for simply-supported sandwich structures to minimize the thermally-induced deformation of composite skins and to maximize the natural frequency and the modal loss factor of sandwich structures is conducted by using a differential evolution in which lamination parameters are used as intermediate design variables. The effect of laminate configuration of composite skins on the thermally-induced deformation characteristic for composite skins and that on the vibration and damping characteristics for sandwich structures are examined through the optimization results.

Tower concentrator solar thermal power plants generate electricity by concentrating sunlight using many reflectors called heliostats. The heliostat has a sun-tracking function, which reflects sunlight onto the tower at the appropriate elevation and azimuth angles according to the sun's orbit. Therefore, the position of the heliostat is important for tower-concentrating solar thermal installations. A hybrid system of solar thermal and photovoltaic power generation is possible by attaching optical multilayer films that can control the spectrum of sunlight to solar cell arrays and using them as heliostats in tower-concentrating solar thermal power generation. This study proposes a heliostat placement method based on the radial staggered method, considering the tower location. The heliostat arrangement is optimized by a genetic algorithm to maximize the total annual light received of the heliostat and tower collector. As a result, all heliostats are placed on the site taking into account the site geometry and tower locations to maximize the total annual light received throughout the facility.

This paper analyzes radiation heat transfer in a solar collector with wavelength dependence for hybridization of photovoltaic and solar thermal power generation. Optical multilayer films with separate transmission and reflection wavelength regions are attached to a solar panel to realize the hybridization. The wavelength dependence of the reflectance of the multilayer, the absorptivity of the heat collector, and the wavelength dependence of the reflected light concerning the solar orbit are implemented. The unsteady temperature distribution of the heat collector is analyzed by coupled light ray tracing analysis and radiation heat transfer analysis. The temperature of the heat collector is affected by the wavelength limitation when the optical multilayer is attached to the solar cell. It was also confirmed that the temperature of the collector area increased even when the wavelength range is divided by the optical multilayer films.

Fractal is a special kind of geometric shape characterized by features such as self-similarity and high complexity. Fractal has a fractal dimension that is different from the traditional topological dimension, which can be used to reveal the complexity of fractal. The metro is an important part of modern urban public transportation, but there is a lack of an intuitive and quantitative tool for optimizing it. We noticed that metro lines are rugged, complex and statistically self-similar. They can be regarded as fractal so that we can quantify their complexity. In this paper, we calculated fractal dimension (d) of metro lines for 27 provincial capital cities with metros in China mainland, and collected data on GDP (e), population (p), city area (s) and population density (ρ = p/s) of these cities to explore the relationship between fractal dimension and these variables. Correlation analysis shows that d has a significant linear positive correlation with e and p, with Pearson's correlation coefficients of 0.8578 and 0.8227 respectively. d has a non-significant linear correlation with s (r_sd = 0.0895), but a significant linear correlation with ρ (r_ρd = 0.6544). Based on the linear correlation of d with p, e and ρ, we performed a multiple linear regression to obtain this equation (d = 1.0436 + 3.2612 × 10⁻⁵e + 3.2767 × 10⁻⁵p + 8.6432 × 10⁻²(p/s) ). Substituting the e, p and s in the equation to calculate theoretical fractal dimension and comparing it with actual fractal dimension, we can evaluate the fitness of current metro lines, which can guide the optimization of metro lines.

This study aimed to enhance the buckling characteristics of composite materials with variable axis stiffness by optimally designing reinforced fiber shapes using the Sequential Approximate Optimization (SAO) method. The fiber shapes were defined by the contours of surfaces, with the defining parameters treated as design variables. The optimization employed Radial Basis Function (RBF) to construct response surfaces, improving accuracy through the sequential addition of appropriate search points, thereby reducing the number of evaluations of the objective function. The results confirmed that under specific boundary conditions, composite materials with optimally curved fiber shapes exhibited significantly higher buckling load factors compared to traditional linear fiber configurations.

Aim of this study is implementing a more precise method for function approximation when dealing with large-scale problems. In this study, we conducted experiments by using Golinski’s Speed Reducer (GSR) benchmark problem (7 design variables). In Convolutional Radial Basis Function Network (CRBFN), with the increasing of the convolution times, the radius of basis centers appear an accelerating upward trend, while the result of evaluation criterion for each iteration shows a slowing downward trend. For they have opposite characteristics, we attempted to use convolutional ways to combine them together. In this experiment, we used 520 sets of data. Each data set includes the design variables, the corresponding function values, and the constraints on the design variables. We used these data sets to train CRBFN and our network to approximate the target function, and the results proved our method was always better.

In practical engineering problems, there are more and more requirements for efficient method that can solve large scale problems. In this study, large scale means the number of design variables are more than 15. Particle Swarm Optimization (PSO) is one of choices that we do not need sensitivity and also it works well even in multi-peaks problems. However, its efficiency is only limited to small scale problems. Most likely number of design variables are less than 6. When the number of design variables raises there are more possibilities to fall into local optimum and PSO cannot get out from there. By its nature, if one achieved to become global best, its velocity becomes close to zero and fall into local optimum. In this study, Agent PSO is used as basic. Agent PSO means that each parameter settings have different characteristics, such as early convergence, middle convergence and late convergence. To avoid early convergence situations, usually we need a new global best to have velocity. In this study, grouping is proposed. It means that we divide population into several groups. Each group has its own global best. And when there are no improvements of personal bests for several iterations, it will look up the other groups global. As for explosions, they are something like mutation in Genetic Algorithms. In this study, two different ways are proposed. If velocities of individuals those who are early convergence, we gave specific number of design variables to random numbers. In several generations, they tend to get close to global bests for their groups and have a chance to get out from local optimum. Another explosion is for those groups which had worst global best, some individuals in that group are given randomly between global best and itself. Effectiveness of the proposed method are given through both Rosenbrock function and Rastrigin function for large scale problems such as 128, 64, 32 and 16 design variables to less than 1.0e-11 order.

In the vehicle development, it is important to solve feasible design sets where all multidisciplinary constraints are satisfied. The set-based design method has been proposed as an effective approach for this purpose. Within this framework, Differential Evolution Based Adaptive Sampling (DEBAS) has been introduced as a technique for exploring feasible regions. However, DEBAS accumulates found feasible solutions, leading to increasing computational costs for calculating the nearest neighbor distance as the number of feasible solutions grows. Consequently, obtaining a sufficient number of feasible solutions within a practical computational time frame can sometimes be difficult. This paper proposes a new method, Differential Evolution Based Adaptive Sampling with Finite Mixture Model (DEBAS-FMM), which applies a finite mixture model to the objective function. This approach aims to mitigate the increase in computational costs associated with the growing number of known feasible solutions, thereby facilitating the derivation of feasible regions more efficiently.

Large UAVs are equipped with geodetic systems such as GPS and various sensors capable of autonomous flight. On the other hand, it is difficult for small UAVs to fly autonomously because it is difficult to acquire location information for self-position estimation. In this study, we propose a method to fly a small UAV to its destination and generate a return trajectory back to its base. 3D point cloud data is used to identify obstacles in the flight space. The UAV flies from the starting point to the target point and back to the starting point by avoiding the obstacles. The trajectory generation problem is formulated as a multi-objective optimization problem. The effectiveness of the proposed method is demonstrated from the generation results of several round-trip trajectories.

In the field of Kansei Engineering, there are many studies on methods for designing product aesthetics by analyzing the relationship between customers' preferences for products (like/dislike) and their design using methods such as Hayashi's quantification methods, rough set theory and artificial neural networks (ANN). Rough set theory is the method that extracts decision rules that explains the relationships between decision and condition attributes from the information of the objects having multiple attributes. In the field of Kansei Engineering, customer’s preferences are defined as decision attributes, while aesthetic elements that make up a product are defined as condition attributes, and decision rules that explain the relationships between them are obtained from the results of questionnaire on existing products. Since the rough set theory is suitable for analyzing uncertain and incomplete information such as Kansei evaluation, it is used in many design methods. In this research, in order to further improve the usefulness of the rough set theory in Kansei engineering, a method for analyzing the relationship between customer preferences and product design is investigated, a method for analyzing the relationship between customer preference and product design based on generative AI, which has attracted attention in recent years, and rough sets is investigated.

Motion in industrial products contributes to both functional and aesthetic qualities, and stimulates user curiosity and interest, thereby encouraging the formation of attachment to the product. In this study, we deduced that the optimal prediction error from typical motion leads to interest, and that fluctuations in prediction error arouse curiosity. Using the free energy principle, a unifying principle of the brain, we formulated that fluctuations in prediction error encourage curiosity and decreases in prediction error arouse interest. We also developed a system to generate motion by controlling prediction error, and experimentally confirmed that the evaluation of interest is convex upward with respect to prediction error and that fluctuations in prediction error increase curiosity.

This paper presents a methodology for quality assurance of KANDO quality linked to human emotion. Surprise, which is a part of KANDO (a positive experience with surprise, which occurs when the experience is greater than the past experience), is used as the subject of the quality assurance. The methodology is developed in the context of Surprise quality assurance in the product planning of a design award-winning stationery product. The QOI (Question Of Interest) and COU (Context of Use) of ASMEV&V40 are defined. The QOI is defined as "whether product planning can guarantee surprise as an emotional quality". The COU of the QOI is defined as "new business planning in a listed stationery manufacturer". The UQ (Uncertainty Quantification) of Surprise obtained through the experiment is propagated to the design and corporate context in which the COU is detailed. Uncertainty propagation uses Monte Carlo simulation and a fuzzy approach. Here, the risk is assessed as the product of quality assurance failures and the losses resulting from the failures. Based on the ASME V&V40, a set of methodological plausibility is then demonstrated. As a result, the quality of Surprise can be assessed as a risk with a range of variability. The results also suggest that the proposed methodology can guarantee the required credibility based on the ASMEV&V40 framework.

Creative exploration is essential in the design of new products. A novel idea, if adopted, can be the launching point for creative inquiry in design, but it requires that the designer accept the idea. Therefore, we considered the designer's acceptability of the idea to be important for creative exploration. In addition, reframing is a cognitive process that generates creativity. In this study, we construct a mathematical model of sensitivity in innovation of meaning by means of reframing, using the framework of the engineering design process that considers structure, behavior, and function, active inference, and the inquiry cycle model of emotion mechanics. Based on the analysis results of the mathematical model, we mathematically explained the designer's acceptability of ideas, and derive the conditions for presenting information that will become ideas that encourage creative inquiry. Furthermore, by analyzing the acceptability of the idea, we found that a smaller idea precision increases the acceptable novelty and curiosity of an idea. Finally, we proposed a method of application such as evaluation and output of ideas based on the derived conditions.

In classical ballet, both technical skill and expressive ability are essential. Conveying emotions, stories, and other elements effectively and impressively to the audience holds significant importance. However, due to the artistic elements integrated with the technical movements, objective evaluation is challenging. This study investigates an objective evaluation method using affective engineering to address the vague and abstract expressions such as "beautiful" and "graceful" commonly used to evaluate the characteristics of body movements in classical ballet. While factors such as background, music, and costumes also significantly impact the overall impression, this study focuses on evaluating skeletal movements. Therefore, OpenPose was used to detect human joint points and generate the BODY_25 model for classical ballet movements. As a result of conducting a survey via Google Forms, responses were received from a total of 50 individuals. Then, rough set theory was applied to analyze the relationship between ballet body movements and impressions.

Recently, product design considering the concept of universal design aimed at satisfying various users has attracted attention required from the social background such as owing to aging and internationalization. To realize universal design, it is necessary to reflect the corresponding the physical characteristics of diverse users in design values, and to consider the physical burden of the user when utilizing the product. However, a general approach to such design has yet to be established. Therefore, this paper proposes a universal design method considering physical burden through motion analysis using a musculoskeletal model, with the aim of designing a product that is easy to use for as many users as possible. Specifically, by expressing the user's physical characteristics in terms of range values and deriving a common range solution that satisfies as many users as possible, we derive a universal design solution that enables as many users as possible to use the product with as little physical load as possible. Moreover, the usefulness of the proposed method is shown by applying it to the design of a display shelf.

In an early design stage, it is critically important to identify latent needs and appropriately fit them into the product. This paper defines the latent needs as those that users recognize as important but cannot explicitly articulate beforehand. However, there is no clear answer to the question of how to identify these latent needs, which remains a topic of academic interest. This study hypothesizes that while users may not be able to describe latent needs in advance, they can evaluate and express desires for additional features when confronted with similar functionalities. To test this hypothesis, we developed a prototype of a sleep-assist robot aimed at preventing sudden infant death syndrome (SIDS) in settings like daycare centers. By having prospective users interact with the prototype, we aim to identify new latent needs that may not be apparent through traditional methods.

In product design work, it is important to study the satisfaction of requirements such as performance and prevention of recurrence/prevention of failure. However, past knowledge may be underutilized, leading to omissions in studies. Recently, knowledge graphs have attracted attention as a way to digitally represent knowledge on computers. In this study, a use case for knowledge graphs in design work was developed, along with a technique for automatically generating knowledge graphs from document information using natural language processing. This technique uses generative AI and CoT (Chain of Thought) to extract knowledge structures step by step. In this paper, we report the results of applying the technique to failure knowledge registered in the "Failure Knowledge DB (https://shippai.org/)". In particular, it features (1) sentence proofreading, (2) classification of sentences into states or failure events, (3) recognition of related interstates and state-events, and (4) diagramming in PlantUML format. Its effectiveness was confirmed by confirming that the state transitions of failure can be extracted and visualized as a knowledge graph.

We are working on improving operational efficiency through DX (Digital Transformation) focusing on the design work in the railway industry to cope with the decrease in the working population. As part of this effort, we have started developing a technology for automatically extracting keywords from documents associated with the information that designers are seeking. In this report, we propose a framework for automatically extracting keywords using natural language processing technologies such as Named Entity Recognition (NER) and TF-IDF method, and the relationships between parts based on BOM (Bill of Materials) information and have developed its prototype. We adopted BERT as the NER technology, and by adding learning specialized words, we have improved the extraction accuracy. Furthermore, we developed the function that estimates which key part each extracted part most relates to by using BOM data and displays the relationship network. We confirmed that this prototype efficiently extracts and visualize keywords from design documents.

In recent years, advances in information technology and large-scale language models have led to a great deal of research on design support through language information processing. In the field of design studies, it is considered effective to divide design problems into layers of structure, behavior, and function, while the concept of user experience has become important in recent years, mainly in the field of design. Given this background, this study proposes a classification of vocabulary and concepts in an electronic concept dictionary into structure, behavior, function, and user experience. This paper reports the results of a study on the effectiveness of introducing the classification into design support by means of language information processing.

In recent years, 3D Computer-Aided Design (3D CAD) has become an essential tool in design, and the need for digital design has increased. While CAD education is provided in universities and companied, there exists a significant disparity in proficiency levels among individuals. Although studies have been conducted to evaluate learners’ operational skills and proficiency, there has been no detailed analysis of the design process, leaving the specific areas where learners struggle and the underlying causes unclear. To address this issue, this study proposes the Digital Transformation (DX) of CAD education with the aim of enhancing its effectiveness and efficiency. By utilizing computers, we collected command logs from 113 learners and extracted command sequences where errors occurred. Based on the operation content, we performed clustering of command sequences before Undo command to accurately grasp the frequency of errors. For each error, we calculated the variation in occurrence frequency among learners’ command logs and the percentage of errors that were corrected using Undo command. This information can be used to estimate the cause of errors and identify mistakes that require individualized attention.

As products become large-scale and complex, manufacturing industry faces problems in communication between workers. Among the information to be communicated, design rationale, such as reasons for decided design and the process of consideration leading to the decisions, is rarely expressed in shared design data. One of the reasons for this is that there is no defined structuring method for the design rationale, such as CAD. In this study, two approaches were used to structure the design rationale toward reusability. The first is that the design rationale is captured in a dynamic model of the design process and a static model of the design object. Modelling the entity and function that is the object of the design operation can be used when selecting what to refer to from the stored design rationales. The second is an extension of DRed(Design Rationale editor), one of the design structuring methods. DRed captures the rationale for the evaluation of design proposals but does not capture the rationale for the ideation of design proposals. However, this study considered that both the ideation and evaluation of design proposals should be recorded. Because the rationale for ideation needs to store for data, assuming that intelligent CAD and generative AI will be able to perform automated design in the future. The proposed model reflecting the two approaches was developed. In this paper, the model is used in a design experiment and a discussion of the result is presented.

In this study, we extend the functional modeling by conversion of matter, energy, and information to functions and user experiences that describes not only physical phenomena but also human physical and psychological characteristics as a conversion model using computational physical quantity representations and linguistic expressions, Based on the similarity between the physical quantity representation and the linguistic expression, we improve the method and software technology to generate design ideas for new functions and user experiences by systematically exploring possible combinations through the connectivity of the transformation model elements. In this study, we will test whether new ideas can be conceived from diagrams generated using a concrete product as an example.

This study examines the ideation capabilities of engineering students, design students, and generative AI, focusing on diversity and specialty in their ideas. The experiment involved presenting each group with various social challenges and asking them to generate short written responses. These responses were analyzed using language information processing techniques to measure diversity—defined as the range of different fields represented—and specialty, which refers to the depth of expertise in a specific area. The results confirmed that generative AI produced the most diverse ideas, followed by design students, and then engineering students. In terms of specialty, engineering students generated the most specialized ideas, especially when explicitly asked to leverage their technical expertise. These findings provide valuable insights into the strengths of each group, suggesting that educational approaches can be tailored to enhance both creativity and problem-solving skills. Future studies will explore additional factors such as abstraction and novelty to gain a deeper understanding of ideation characteristics.

Visually impaired person uses a white cane or Braille blocks to confirm their safety when walking, but this is insufficient. There is the problem of not placing the foot on the ground point where safety is confirmed by the white cane, and the problem of not being able to recognize obstacles in the air. We developed a wearable system that enhances visual information for the visually impaired by combining multiple distance sensors. The system is worn around the neck and can be easy to use with a white cane. The sensor can keep horizontality by a stabilizer against shaking during walking. To measure the performance of the sensor, we performed measurement experiments conducted using obstacles assume during walking. As a result, the sensor was able to capture the shape characteristics of a person, a truck side mirror. In the future, we will also conduct measurement experiments with other obstacles. Future developments will consider method of posting hazard information and guidance on obstacle avoidance.

In recent years, advances in industry have improved technological capabilities, leading to increased functionality and complexity in products. During maintenance of complex products, it is rare to have work instructions for all components, and the absence of such instructions has led to occupational accidents, including those endangering lives. Consequently, in tasks such as disassembly for complex machinery maintenance, workers often need to determine the disassembly sequence themselves. Therefore, this study focuses on the eye movements of workers during product disassembly to discover insights that facilitate smooth disassembly operations. To analyze the characteristics of eye movements relevant to determining disassembly sequences, experiments were conducted where workers wore eye trackers and disassembled a model car with the objective of extracting specific components. Quantitative and qualitative analysis revealed a tendency for workers to begin disassembling from components they focused on, and to proceed to remove parts similar to those dealt with in the preceding steps.

In recent years, the number of workers engaged in railway maintenance has decreased. Line maintenance requires a high level of knowledge, extensive experience, and technical skills. Furthermore, oppressive working conditions and a declining birthrate and an aging population have contributed to the decline in workers. Solving these manpower shortages can be effectively achieved by the introduction of robots that automate and enhance the efficiency of maintenance tasks. The modular robot developed in this study is lightweight at 0.2kg per one module and portable, running on a single rail with six wheels in a straddle-top style. It can recognize objects three-dimensionally by combining 4x4 matrix distance data from an ToF sensor and images from a small camera mounted on the upper front of the body. Additionally, an encoder is used to determine the robot's position and inform the operator, aiming to enhance line maintenance work efficacy. A static measurement experiment was conducted by placing a plastic cube of 0.1 m per side at 0.1 m from the rail without moving the body. Afterwards, static experiments were conducted by gradually moving the cube away, up to a maximum distance of 0.7 m. As a result, obstacle detection was confirmed. On the other hand, during a same experiment with the robot moving, continuous data could not be obtained due to the data processing load. The experiment results indicate that data can be observed and processed by enhancing the processing capacity and speed while reducing the running speed of the robot. These findings are expected to reduce the burden of line maintenance work and facilitate safer, more efficient railway maintenance.