Product modularization is an architecture design scheme for a product to manage its complexity as a system. The design methods for modularization often employ mathematical operations to achieve better modularization. Although multiple objectives result in various alternatives, identifying the superior one best suited for their tradeoff relationships and ambiguity in their interpretation is challenging. Moreover, the decision-maker is required to explore the design space and determine the preferred solution with a rational reason. This study proposes a decision support method for product modularization based on the design structure matrix (DSM). The method is directed by the hierarchical clustering of alternatives into a dendrogram, and it facilitates the interpretation of alternatives over a decision bifurcation diagram. It is assumed that an optimal solution can be represented as a piling-up of smaller pieces, which correspond to building blocks in the combinatorial solution space. The distance of alternatives was surveyed using schema representation defined in various levels. Further, the hierarchical clustering technique structures the design space of alternatives into branches and organizes the essential parts, identified as building blocks, into a tree-shaped graph. Consequently, such a structure is crucial to revealing the leverage points in modularity and facilitates the exploration of product modularization. The proposed method was applied to an air-conditioner unit to demonstrate and ascertain its validity and effectiveness.

In selective laser melting (SLM) metal powder is completely melted using a high-power laser beam of 200-400 W to form metallic parts which are almost completely dense and do not require post processing. To reduce the effective Young's modulus of fabricated parts by SLM to that of human bone, strut structures with interconnected pores of mm size are usually adopted for bone implants. In contrast, parts fabricated by SLM or selective laser sintering (SLS) under laser power below about 50 W are not fully dense and contain micropores, resulting in the reduction of the part’s Young's modulus. However, low-power laser SLM/SLS techniques have yet to be extensively studied. In this study, we examine the densities, effective Young's modulus, and metal structure of parts fabricated by low laser power SLM from pure Ti powder under various experimental conditions. The effective Young’s moduli of as-sintered cubic specimens were a few GPa independent of energy density. After applying normal loads, the effective Young’s moduli increased in various rates, according to energy density. At a normal load of 10kN, the effective Young’s moduli were divided into three groups, 10, 40 and 100 GPa. The relative density ranged from 50 to 75% and cluster size from 70 to 330 μm depending on the energy density.

In a production system using advance demand information, the advance demand information is sent to an upstream company a certain period of time before production, and the firm-order information is sent just before production. Unfortunately, upstream companies are frequently burdened with excess inventory and late deliveries. Therefore, it is necessary to analyze the effect of the variable factors, such as demand fluctuation and the gap between the advance demand information and the firm-order information. The model was based on the following three steps. First, considering a supply chain consisting of four companies, suppliers can be classified into three types, according to the timing of the firm order and production. Second, the supply chain consists of a combination of three types of suppliers, depending on the timing of the sending firm order. Finally, the production-activity structure for each type of supplier was modeled. Then, they were formulated and implemented as an analysis simulator. Using this simulator, several case studies were investigated. As a result, it was confirmed that the bullwhip effect was increased by the sudden fluctuation of demand, especially when the demand increased. In addition, the supplier whose daily production numbers and stock quantity fluctuated was drastically changed by the gap between the advance demand information and the firm order was identified.

As a solution for sequence optimization, and to obtain a solution that is strong robustness to variations, a multi-objective robust design optimization (MORDO) and robust genetic algorithm (GA) are applied to a multi-objective optimization problem. As a result, it is found that the MORDO is a good method for obtaining a solution with strong robustness to condition variations, even in sequence problem optimization. Moreover, it is confirmed that the robust GA is advantageous as in regards to the calculation time. As a method for making the best use of these advantages, a hybrid method of the MORDO and robust GA is proposed and verified, and an optimum solution with the same robustness as that from the MORDO is obtained without huge loss of calculation time.

This study aims to optimize the process parameters for PBF-LB of Inconel 718 using the surface texture and density of the as-built specimen to fabricate high-quality products. In order to achieve this aim, the melting and solidification phenomena of the melt pool were examined using a high-speed camera, and the process map of laser power and scan speed was constructed by evaluating the surface texture and density of the as-built specimen to optimize the process parameters. In addition, the mechanical properties of the as-built specimens fabricated at the optimum process parameter derived from the process map were examined. Consequently, it was found that the optimum process parameter should be determined by consideration of the melting and solidification phenomena because the phenomena such as spattering and instability of the melt pool change with the process parameters. It was also found that the morphology of the track is affected by the spot size of the laser beam. It was revealed that the similarity of the process maps evaluated by surface texture and density of the specimens results in the correlation between the surface texture and the density. The optimum laser power and scan speed can be determined using the process maps evaluated by the surface texture and density of the specimens. The as-built specimen fabricated at the optimum process parameter showed anisotropic tensile behavior, and the tensile strength of the transverse specimen was higher than that of the longitudinal specimen. The tensile properties of the as-built specimen fabricated in this work were similar to those of the as-built specimen reported by the other works.

In the early stage of dynamic system development which has a multi-disciplinary and hierarchical structure, system requirements need to be cascaded down to target values of each component so that engineers can collaborate efficiently and concurrently. The purpose of this paper is to propose a novel set-based concurrent engineering method for a dynamic system by using machine learning. In the practice of the target setting study for concurrent engineering, both hierarchical simulations between system and component level and a solution to solve inverse problems using the simulation are required. The proposed method composes of two machine learning methods that satisfy these requirements. The first one is physics-informed long short-term memory (PI-LSTM) which enhances the mechanical modeling of component behavior. By applying the proposed PI-LSTM to where mechanical modeling is difficult, the adaptive range of mechanical modeling can be expanded. The PI-LSTM surrogate the dynamic behavior of the component model and can be used inside the system-level simulation. The other one is Bayesian active learning (BAL) applied to inverse problems to solve feasible regions where all system requirements are satisfied. In the proposed BAL, Gaussian process models are trained from the system-level simulation, and an acquisition function is evaluated and maximized to generate new sampling candidates. The set-based design using BAL has an advantage in the decoupling ability of design problems because feasible regions of each discipline sub-problem can be studied concurrently. To show the effectiveness of the proposed method, a numerical example of a vehicle design problem which has a hierarchical structure is demonstrated.

Stock-type warehouses at production sites and supply chains face a demand for quick and timely delivery of materials and products. However, the mainstream method of delivery work is still manual order picking and improving the work efficiency is a crucial issue to achieve smart logistics and/or factory. Therefore, in this study, we propose an optimization method using order batching (OB) for multi-order picking. In addition, a method is also proposed to optimize storage location assignment (SLA) to reduce the travel path of pickers. In this case, the number of location pairs to change is limited according to actual operations. Furthermore, numerical experiments validate the co-optimization method for coupling OB and SLA. As the knowledge from the numerical experiments, the co-optimization coupled by OB and SLA shows that the control and execution of the local optimization techniques according to the order characteristics is efficient and effective.

During push switch operation, tactile feedback is crucial in determining whether the operation will be successful. However, to enhance the tactile feedback and ensure usability, the actuation force should not increase. Herein, a method to improve tactile feedback during switch operation by designing the operating sound is proposed. In the first experiment, to investigate the acoustic characteristics of high feedback operating sound, participants evaluated recorded commercial push switch operating sounds through headphones using the semantic differential method. Results showed that the operating sound with a short decay time and high loudness value in the high-frequency band exhibited a highly intelligible impression. Furthermore, the push switch operation exhibited two sounds (push and release), whose contribution to the impression was determined based on their loudness. In the second experiment, participants evaluated their impressions when operating push switches, which comprised various combinations of operating sound volume and actuation force. Results showed that highly intelligible sound improved the tactile feedback. Moreover, it was found that considering the balance between the operating sound volume and actuation force is necessary to improve the switch operation feedback through the operating sound. These findings can aid in enhancing the feedback of devices that require switches with low actuation force due to design constraints.

Ultraprecision cutting used for creating highly precise components and particularly micro-shapes requires ultraprecision machine tools. However, these machine tools are operated under restricted cutting conditions such as feed rate and depth of cut, resulting in low productivity. Therefore, this study deals with this problem by devising a novel ultraprecision cutting system with automation of the workpiece setting operation. In the devised system, the workpiece is roughly machined by an ordinary machine tool. After that, an industrial robot transfers the workpiece to an ultraprecision machine tool to complete the rest of the machining process. The proposed machining system overcomes two types of errors. The first is the shape error due to rough machining on the ordinary machine tool. It is practically difficult to create the expected shape exactly in rough machining, and therefore the shape error of the roughly machined workpiece is affecting the finish machining directly. The other is the setting errors due to the transfer of the workpiece to the ultraprecision machine tool from the ordinary machine tool. The differences from the ideal workpiece position and orientation are detected to identify the setting errors. Thus, the roughly machined workpiece is scanned to derive an error map by on-machine measurement. Additional tool paths are generated for semi-finish machining, and the error map helps to eliminate the form errors induced by rough machining. The NC program for finish machining is modified to compensate the identified workpiece setting errors. Finally, finish machining can be conducted on the roughly machined workpiece. Hence, the removal volume in finish machining on an ultraprecision machine tool is reduced, and the time required for the whole machining process is expected to be shorter. The experimental results confirm that the developed ultraprecision cutting system contributes to the automated workpiece setting operation to create micro-shapes with high accuracy.

Work-related musculoskeletal disorders are the most frequent health issues, with awkward posture being one of the risk factors. Observational methods are often used in the manufacturing industry to analyze work postures in production fields. In this study, we present a straightforward technique for evaluating work postures utilizing the Ovako working posture analysis system (OWAS). The proposed technique calculates OWAS-based posture codes by manually acquiring employees’ two-dimensional (2D) joint coordinates on the work image and inputting these coordinates into advanced machine learning models. Experiments were conducted to extract three-dimensional (3D) joint coordinates in the global coordinate system in the OWAS-based postures to develop machine learning models. Furthermore, the resulting 3D coordinates were converted to 2D joint coordinates in the camera image coordinate system using the direct linear transformation (DLT) method. The 2D joint coordinates and accompanying OWAS posture codes were utilized as training data to build machine learning models using the support vector machine algorithm. Cross validation confirmed the agreement rate of the OWAS action category (AC) by more than 80%, according to the experimental results.

The purpose of this research is to verify the method in the elicitation of latent needs from consumer needs by conducting the working prototype-based interview and collecting raw data which is responses from the consumer. Latent needs are those that many consumers recognize as important in a final product but do not or cannot articulate in advance. The challenge in identifying latent needs is finding the method to elicit from consumers the needs which are not addressed by any inventors yet in the present market but would delight consumers if delivered tomorrow. After conducting a consumer feedback questionnaire via the internet, a working prototype was created. The working prototype then was used as material to prepare presentation slides. The first presentation slides were focused on the background problems and ideas for the solutions while the second presentation slides provided consumers with a prototype and story of the product that was believed would be one of the solutions to the problems. Based on the results, latent needs interpreted from interviewees’ responses and the categories of the needs obtained from the Prototype-based interviews are more than from the Problem-based interview. The latent needs that we were able to obtain from this research were for example, “The device is able to detect small changes in a child while changing a diaper” and “The device is able to detect small changes in a child while watching he/she sleeping” which could lead into the prevention of unwanted incident such as sudden infant death syndrome (SIDS). This supports our assumption that showing working prototype based materials with story descriptions can be effective in uncovering potential latent needs. However, due to the COVID-19 pandemic, we were unable to give the interviewees chances to touch and look closely at the working prototype therefore latent needs possibly gained from this experience are still uncovered. Although there are still limitations in our findings, the method that we proposed is able to support discovering latent needs in future.

In logistics warehouses, customer needs have diversified and product life cycles have shortened. Hence, it is necessary to deliver various types products to retailers in a short period of time while responding to fluctuating demand. It is important to increase the efficiency of the order picking in warehouses. When multiple workers pick simultaneously, psychological stress develops due to the presence of other workers. When other workers pick on the storage rack, it is not possible to work on the same storage rack, leading to psychological waiting stress until picking of other workers is completed. In order picking, it is important to consider not only the work time but also the stress of workers. However, no research considers the psychological stress in order picking. We used multi-agent simulation and proposed a storage assignment change plan under fluctuating demand considering work time and psychological stress. We determined the optimal pattern of storage assignment change that minimizes work time or psychological stress according to the fluctuating demand. Assignment of work areas based on the optimal pattern of storage assignment changes further reduced work time and stress. By determining the reduction of either work time or stress and selecting the pattern of storage assignment change and assignment of work areas, flexible and appropriate picking operations can be performed according to the daily situation.

In face-to-face communication, talkers perceive various affects and emotions of other talkers as the eye-impressions such as highlight or dilated pupil. In the previous studies, focusing on the eye-impressions in social robots, various expression methods such as changing the LED color or blinking the LED were proposed and developed. However, it is difficult to generate the affective expressions with the highlight that has an optical reflection characteristic on the surface of the eyeball by using the LED. Therefore, in order to enhance the eye-impressions of social robots, it is expected to develop an expressible method with the affective highlight. In this study, we proposed an affective highlight to enhance the eye-impressions of social robots. This method introduces a virtual lighting in virtual space and can generate rich affective expressions by controlling various parameters of the virtual lighting such as brightness or colors. In addition, we developed a speech-driven pupil response system that applied the affective highlight to the pupil response interface which has been developed. This system can generate the affective expressions by the dilation of pupil response that is synchronized with the talker’s speech as well as the highlight expression with talker’s intentional input. The effectiveness of the developed system with the affective highlight was demonstrated by sensory evaluations.

In the manufacturing industry, various types of machine tools are widely used by many engineers. Some inexperienced users of machine tools, such as students and trainees, operate machine tools when they are training to acquire skills. However, machine tools have mechanisms that rotate at high speed and exert large forces, which can be dangerous for inexperienced users, and therefore safe training of personnel is required. Virtual reality (VR) is inexpensive compared to machine tools and can provide an immersive training environment for a large number of personnel. In this study, we developed a training system for machining training in a virtual space for a desktop milling machine tool. Based on the input NC program, the movement of each axis and the rotation of the main axis are represented, associated with the control panel by C# script. As a training function, interference between the tool and workpiece, hand contact with the tool, and overrun of each axis are determined by interference checks for each part, and an interface is created that displays warnings on the user's line of sight. In addition, the machined shape is reproduced by using a function called “stencil” in Unity. Furthermore, in order to realize the digital twin, the behavior of the actual machine and that of the virtual machine can be synchronized using TCP/IP, and the behavior of the table of the actual machine tool can be monitored.

Platform-based Product Development (PPD) approach is an effective way to achieve mass customization. The product platform configuration in PPD is to determine the number and composition of platforms for a product family. Different types of design strategy have often been found to satisfy product requirements, i.e., the matching-designed with higher platform development cost and lower customization cost, the over-designed or under-designed platforms with different performance in these two types of cost. Traditional research balances the cost trade-off within the design domain, and few studies include the procurement cost from suppliers. Involving the supplier selection requirements into product platform configuration was examined in this study with focus on module-based platforms. The proposed model contains development cost, sourcing cost, and customization cost, under given product architecture and given number of modules and associated options. We developed a solving algorithm applying the linearization method with Gurobi solver to solve the proposed nonlinear MIP (mixed integer programming) model. Numerical experiments are carried out and sensitivity analyses are performed to justify the proposed model. The results show that the optimal combination of suppliers and the optimal number of platforms depends on the given parameters, such as various cost parameters included in the model, product demand and lifetime. Sensitivity analyses show that the optimal number of product platforms will decrease as the development cost increases, while the number of platforms increases as the customization cost increases. The over-designed platform is more prevalent in the presence of high development cost and high customization cost. In contrast, the matching-designed platform is more suitable for low development cost and high customization cost.

In this study, we propose a method for structurally understanding the differences and similarities among the alternatives of a bi-objective permutation scheduling problem. Although metaheuristics-based optimization technique can exploit Pareto-optimal solutions, it is still difficult for decision-makers to identify the preferred solution among them. In scheduling problems, a distinctive category of combinatorial optimization problems, situations frequently arise in which objective function values are similar despite having different combinatorial structures. Effective decision-making requires a method that focuses on the internal information of the alternatives rather than a simple comparison study of the objective functions. This study applies the dendrogram-directed design space exploration technique to a permutation flow-shop scheduling problem by arranging its modeling scheme. The technique systematically organizes the structure of alternatives and systematizes their features into a tree-shaped graph to provide decision-makers with valuable insights. Clarifying the leverage points of the combinatorial problem with a graph reduces the cognitive burden on decision-makers. As the dendrogram systematically bifurcates the alternatives, the exclusive OR of a pair of clusters indicates their characteristics and differences. A model using the design structure matrix is proposed to abstract the characteristics of alternatives for scheduling problems, and a method for classifying and structuring the analysis of alternatives using it is proposed. Decision-makers can explore the characteristics of leverage points by referring to their position in the objective function space. The proposed technique is demonstrated by applying it to a simple benchmark problem.