Quartz Crystal Microbalance (QCM) was introduced into the surface force measurement to investigate the contact mechanics by analyzing resonance frequency and impedance. PDMS (Polydimethylsiloxane) spherical probe and the Au electrode of QCM were brought into contact and separated at a constant withdrawal speed. Considering the mechanical contact model of the experiment, we constructed the equivalent electrical circuit, evaluated the inductance, capacitance and resonant resistance by the impedance analysis, and converted them again to the mechanical quantities (contact mass, contact stiffness and contact damping coefficient). The results showed that contact mass, stiffness and damping coefficient decreased during the separation, which indicates that impedance analysis is one of the practical tools for pursuing contact mechanics.

Articular cartilage is hydrogel, which consists of nano-scaled network structure and water, and shows high lubricity. Although the lubricity has not been well understood, it should be driven by dynamic behavior of the network structure and water. The dynamic behavior of the network structure during friction should be observed at the nano scale. However, the method to visualize the network structure has not been developed. We are developing a friction tester between micro agarose probes under scanning electron microscope visualization. In this study, we fabricated micro agarose probes on the tip of cantilevers. A micro agarose bridge was formed between cantilevers, and cut with the contact of a hot tungsten wire of 10 μm in diameter. A structure that retained water by surface tension was placed near the tip of cantilevers to prevent the agar from drying out while cutting the micro agar bridge. The radius of curvature of agarose probes was 91 μm.

Rotating flows driven by rotating disks exhibit complex three-dimensional structure in closed enclosure. The structure is caused by the secondary flow formed by centrifugal forces near the surface of the rotating disk. However, the details are yet to be known especially in non-axisymmetric enclosure and with an obstacle. In the present study, we focus on the flow in a hard disk drive (HDD) as a model. We investigate the structure of the secondary flow formed in the rotating flow between two parallel disks sharing a central axis in a non-axisymmetric container. Although PIV can measure the flow in the parallel disk cross section, the direct measurement of the axial flow remains difficult. Therefore, the axial velocities were reconstructed by applying mass conservation from the in-plane velocity distributions at different axial heights. We considered a more reliable reconstruction by modifying the calculation scheme. In the process, we utilized higher-order finite difference methods and numerical integrations to examine the effect of increasing the number of intervals and the effect of overlapping of the PIV measurement planes.

We investigated the motion of small particles in a rotating flow driven by corotating disks within an enclosure featuring a partial shroud opening, experimenting flow in a hard disk drive (HDD). In HDDs, levitating particles from mechanical components can damage the disk or magnetic head. Using a refractive index-matched system for clear visualization, we tracked nearly neutrally buoyant particles and analyzed their trajectories in polar coordinates parallel to the disks. Particles initially moved outward due to centrifugal forces, collided with the shroud, and reflected inward at specific angles. Statistical analysis revealed preferred trajectories, with higher passage frequencies at certain positions, highlighting characteristic flow patterns.

It has been reported that alkyl diphenyl ether (ADE) exhibits excellent wear resistance and does not decompose unless nascent surface is exposed. However, under extreme contact conditions, decomposition of ADE was observed when the sliding surfaces wore down. In this study, the goal is to reduce wear and delay decomposition by using sulfur-based lubricants in which sulfur is introduced into the chemical structure of ADE. Using Q-mass, the tribochemical decomposition properties of ADE and sulfur-based lubricants were investigated. ADE decomposed at a sliding distance of 6 km, whereas mono alkyl phenoxathiin (MAPT) and di alkyl phenoxathiin (DAPT) were observed to decompose at a sliding distance of 30 km or more. APTs are thought to have exhibited a long induction period because they produce iron sulfide on sliding surfaces and acquire wear resistance. In addition, DAPT with more side chains showed better wear resistance. This is thought to be because the increase in side chains reduces the shear strength of the interface compared to MAPT.

To understand the decomposition phenomenon of perfluoropolyether lubricants in heat-assisted magnetic recording, we investigated the decomposition of lubricants under high temperature and high pressure in the accelerating aging tests using sealing glass ampoules. Z-15, Z-tetraol, and D-4OH were used as test lubricants in this study. As a result, it was found that PFPE lubricant decomposition was accelerated under high pressure and high temperature compared to that under ambient pressure and high temperature, that the presence or absence of oxygen had no effect, and that decomposition from the terminal hydrocarbon segments was the major factor.

Heat-Assisted Magnetic Recording (HAMR) is expected for one of future high density recording technologies. When the magnetic film is heated at Curie temperature (Tc) around 850K (577 degree Celsius) for FePt system by laser beam via near-field transducer (NFT), multi-layer system, i.e. NFT, lubricant film, carbon overcoat film (COC), heat sink and underlayers, is also heated. Therefore, it is important for design those materials to measure every temperature of layered structure based on thermal degradation, and thermal diffusion around magnetic film. We have developed new measurement system based on surface-enhanced Raman spectroscopy (SERS) using transmission-type plasmonic sensor (TPS) which has a function of chemical analysis for ultra-thin film with atomistic depth resolution. The measurement system can emulate HAMR process because that is equivalent to the HAMR system in physics. Our system is also able to measure temperature of every layer component using anti-Stokes/Stokes intensity ratio and radiation spectrum. We report the temperature profile of layered component in FePt HDD involving effect of gap between the heating device and the recording medium on temperature in laser heating.

In the extraction and separation of stacked paper sheets, various mechanisms utilize the frictional force of rubber rollers. The frictional force of the rubber rollers, which is crucial for separation conditions, decreases due to surface deterioration, adhesion of paper dust, and other contaminants, leading to reduced conveyance speed and separation anomalies. This presentation focuses on a technology for early detection of roller deterioration and abnormalities in a separation mechanism using a drive roller and a driven roller with a torque limiter. Specifically, it demonstrates that by arranging paper detection sensors, the variation in paper conveyance time within a specified section can be minimized, aiding in the early detection of rubber roller deterioration and potential abnormalities.

Inertial navigation in automobiles and other applications requires navigation-grade accelerometers with high accuracy and low cross-axis sensitivity. This paper introduces a miniaturized modularized MEMS differential resonant accelerometer (DRA). The DRA was formed by processing the device layer Si of an SOI wafer using Deep RIE and etching the BOX layer SiO2 using Vapor HF. After dicing, the MEMS chip is die-bonded and wire-bonded to a ceramic package, and then vacuum-sealed. Inertial force causes the proof mass to rotate around a pivot, which generates tensile and compressive stresses in the two resonant beams. As a result, the resonant frequency changes in the opposite direction, which is the operating principle. By reducing the nonlinear vibration and interference of the two resonant beams and performing acceleration and temperature compensation (G-T compensation), a highly accurate bias instability (BI) of 430 nG is achieved. In addition, a cross-axis sensitivity of 0.82% was achieved.

This study reports a method to detect and compensate X-Y mismatches of Q-factors and resonant frequencies in frequency modulation and rate integrating gyroscopes. The mismatches are detected by CW and CCW mode coupling combined with amplitude modulation-demodulation technique. The Q-factor and frequency mismatches could be compensated by amplitudes and phases of of driving signals. A signal processing method to suppress interference between feedback control loops is proposed. Mismatch signals were successfully detected and kept near zero values for more than 20 hrs.

This study aimed to improve the accuracy of predicting mechanical parameters of MEMS devices using Convolutional Neural Networks (CNN). In prior research, a CNN model was trained with sets of photomask images of MEMS devices and their corresponding mechanical parameters, thereby creating a model that could predict these parameters from the images. In this study, we propose a method to enhance the accuracy of the CNN-based approach from previous research by incorporating attention mechanisms into the CNN. By combining spatial attention mechanisms，channel attention mechanisms, and convolutional attention mechanisms, we developed a model that simultaneously predicts spring constants, mass, and resonant frequencies, and conducted predictions on both validation data and actual devices. Experimental results confirmed that the proposed model outperformed existing methods. The introduction of attention mechanisms not only enhanced the CNN's ability to extract local features but also allowed the model to focus on specific regions, thereby increasing its robustness against noise and data omissions.

This study focuses on the control of electrostatic levitation of silicon wafers using deep reinforcement learning (DRL). The aim is to develop a control system that can stably levitate silicon wafers by adjusting the applied voltage. In this research, the state input to the neural network consists of the position and velocity of the levitating silicon wafer, while the action is the applied voltage. A physical engine is used to calculate the wafer's position and velocity under vacuum conditions. The Q-learning algorithm was first implemented, followed by the development of a Deep Q-Network (DQN) to address the challenges of large state and action spaces. Simulations were conducted using silicon wafers of different sizes (2 inch, 4 inch, 6 inch), and the results showed that for larger wafers (4 inch and 6 inch), stable levitation was achieved with specific voltage settings. However, smaller wafers (2 inch) failed to levitate due to the limited voltage range. Additionally, the improvement in control performance with increasing voltage range was demonstrated by simulations using four different applied voltages (0 V, 30 V, 100 V, 500 V). These results demonstrated the effectiveness of electrostatic levitation control using reinforcement learning.

This study investigated the influence of the shape of elastic components that support the electrodes in array-electrode-type electrostatic chucks (ESCs) by comparing two designs: one using coil springs as elastic components, referred to as the spring type, and the other featuring inclined elastic beams, referred to as the beam type. Both types exhibited equivalent maximum electrostatic adhesion forces. However, the spring type demonstrated higher isotropy of adhesion force with respect to the parallelism between the electrode surface and the gripping target surface, while the beam type showed lower isotropy. These results highlight the importance of designing the shape of elastic components and the method of posture control to meet specific requirements, such as adhesion and detachment performance and dimensional constraints, in ESC applications.

In order to test the sensitivity of cells to vibration while maintaining a clean environment, we are developing a magnetic micro oscillator to be incorporated into a cell vibration micro-device in a dish with a lid. Previous studies have shown that the vibration characteristics of the micro oscillator are not appropriate and that friction between the dish and the micro oscillator has a negative effect on vibration. In order to improve these problems, we have redesigned the vibration characteristics of the oscillators, fabricated oscillators with micro protrusions to reduce the friction, and evaluated the vibration characteristics. The resulting displacement of oscillator was increased by micro protrusions. The redesigned magnetic micro oscillator showed sinusoidal oscillations under certain conditions.

In this paper, we propose a photothermally actuated micropump using excitation light from a microscopic observation system. The micropump consists of a chamber with a microheater made of a photosensitive composite, a flow channel for reagents, and a flow channel for biological samples. The photosensitive composite used for the microheater is a mixture of photosensitive resist SU-8 and copper particles. Since the copper particles generate heat due to the photothermal effect, the irradiated microheater exhibits rise of temperature. The air expansion in the chamber caused by the rise of temperature is utilized for pumping. The fabricated microheater demonstrated a maximum temperature rise of 39.3°C, while the fabricated micropump showed pumping volume of 0.58 nL at 36.80 W/cm2 in irradiance.

The purpose of this study is to represent race walking motion in the frequency domain and to elucidate periodic characteristics of race walking motion in the frequency domain. Using power walking as a comparison for race walking, we will examine the changes in the characteristics observed in acceleration of 15 body parts when the pace is varied from 120 to 200 BPM in 10 BPM increments. In the previous report, vertical acceleration was observed. In this report, the characteristics of motion are elucidated using forward and backward acceleration. The analysis revealed that lower leg characteristics differed between race walking and power walking. This result indicates that the leg is the part in which athletes perform the motions required to comply with the rules of race walking.

Ground reaction force (GRF) is a useful parameter with which to evaluate balance in walking, however there is no simple method of obtaining it. Therefore, the objective of this study is to develop a method of estimating GRF that is both simple and accurate. This report proposes an improved approach to estimating vertical GRF, with a focus on improving estimation accuracy. Additionally, it explores generalization of a deep learning model that utilizes computer graphics images as training data. In the previous report, a deep learning model was developed using computer graphics based on motion data measured by one person as training data. However, this approach did not yield highly accurate estimation. The present report aims to improve the accuracy through fine tuning. To this end, a deep learning model was constructed using a dataset consisting of real images of walking as a pre-training, and a dataset consisting of computer graphics images during fine tuning. As a result, there was an improvement in the accuracy of the estimation. The generalization performance was examined by estimating three unlearned participants. The results showed that the utilization of computer graphics and fine tuning improved the generalization performance.

This study investigated the effects of compression socks on body surface vibration and muscle activity during running. Ten healthy male participants in their 20s with exercise experience ran on a treadmill at 12 km/h for 7.5 minutes under two conditions: wearing and not wearing compression socks. Lower leg vibration and electromyography (EMG) were measured during running and muscle hardness was measured before and after exercise. The Power Spectral Density (PSD) analysis showed that wearing compression socks significantly reduced peak PSD of vibration in the X-axis during stance and in the Y-axis during swing. The change in RMS EMG suggested that compression socks may support leg movement and enhance muscle activity efficiency in some participants. Additionally, compression socks suppressed the increase in muscle stiffness after exercise.

The purpose of this study is to establish a decoding technique to estimate sounds heard by humans from fMRI images using deep learning. The sounds we hear in usually have a distinctive timbre. The timbre is determined by the combination of sound pressure levels of the overtones in compound tones, i.e., the frequency spectrum. Previous studies suggested that tonotopy, which shows a specific pattern of activation in the auditory cortex, is influenced by the frequency spectrum. In this report, we focus on tonotopy and estimate timbres from fMRI images using deep learning. Four types of timbres were prepared, and four pitches differing by two timbres each were learned by two classifications to create six classifiers. When the classifiers were used for estimation, the maximum estimation rate for untrained data was 67.22%, and the average estimate rate for the four timbres was 45.31%, far exceeding the chance level of 25.00% for four classifications. This suggests that the proposed estimation method is useful. In addition, issues remained for unlearned data including unlearned pitches.

This paper proposed a method to obtain heartbeat intervals with the higher temporal resolution level of ECG (electrocardiogram) from lower temporal resolution PPG (photoplethysmogram) using a 1DCNN estimation model. First, ECG and PPG were simultaneously recorded. ECG was pre-processed to remove all other components except the R-waves, which indicates the contraction of the cardiac ventricle, or in other words, exact time of heartbeats. Then using PPG as the input data, and the pre-processed ECG, called R-only-ECG, as the teacher data, the 1DCNN estimation model was trained to estimate the R-only-ECG from PPG alone. Finally, using the trained estimation model, higher temporal resolution R-only-ECG can be estimated from lower temporal resolution PPG, thus enable higher temporal accuracy heartbeat intervals be obtained from the estimated R-only-ECG. Both correlation analysis and Bland-Altman analysis were used to evaluate the accuracy and consistency of heartbeat intervals obtained from ECG, PPG and the proposed method. The results verified higher correlation coefficiency and smaller 95% confidence intervals of heartbeat intervals obtained using the proposed method than those obtained from PPG directly. This demonstrated that the proposed model has the potential to improve the accuracy of health monitoring.

In this study, we derived the sound characteristics in a working environment to evaluate their impact, particularly the tempo of background sounds, on task efficiency estimation. In an experiment, physiological data were collected from participants performing a continuous addition task while listening to background sounds, thereby creating a dataset based on the collected data. Using this dataset, machine learning models were developed to estimate the task efficiency and compare the performances. The results demonstrated that the proposed method, which incorporates the tempo of background sounds as an explanatory variable, achieved significantly higher accuracy, interpretability, and stability than conventional methods that rely solely on physiological data. Furthermore, the tempo of the background sound was found to be a highly important explanatory variable, effectively contributing to the improvement of model performance as an external factor representing the working environment. These findings suggest the effectiveness of incorporating the tempo of background sound as a key factor in enhancing the accuracy of task efficiency estimation models for tasks performed while listening to sounds.

In this study, we evaluated changes in mental and physical states before and after viewing seasonal images that symbolized the four seasons of the temperate climate found in Honshu, Japan. The temperature and humidity at the beginning of the experiment were controlled, and a sense of the seasons was evoked through visual images. Biometric information was monitored through finger volume, pulse wave, skin conductance, gazing time, and pupil diameter. Questionnaires were administered for mood and image evaluation before and after the experiment. The results suggest that the effect of decreasing the arousal level increases when the difference between the outdoor ambient temperature and image temperature is large, and the degree of preference is high. In the experiments conducted during the warm season, winter images suggested that the arousal level decreases, and the cold weather experiment suggested that spring, summer, and fall images may have the effect of decreasing the arousal level. The summer images produced a relaxing effect, with no significant difference in pupil diameter reflecting the low preference for summer images and a low arousal reduction effect. This study suggests that visual seasonality may contribute to stress reduction and can be applied to stress management and mental health improvement.

This study aimed to label paintings based on biometric data and generate paintings that elicit specific emotions using machine learning. To construct the dataset, experiments were conducted on eight participants using multi physiological sensors. Following Russell's model of emotion, the arousal-sleepy axis was assessed via skin conductance, while the pleasure-unpleasure axis was evaluated through subjective ratings. Based on these measures, paintings were categorized into four emotional states: arousal-pleasure, arousal-unpleasure, sleepy-pleasure, and sleepy-unpleasure. A generative adversarial network model was then employed to generate paintings corresponding to these emotional states. Viewing experiments were conducted using multi physiological sensors, wearable optical topography, and subjective evaluations to assess the emotional impact of the generated paintings. Skin conductance results suggested that the generated paintings may not effectively induce arousal-related emotions. Additionally, questionnaire responses revealed that paintings in the arousal-unpleasure category significantly conveyed unpleasant emotions, suggesting that machine learning can effectively capture features associated with unpleasant emotions. These findings indicate that machine learning can partially induce specific emotions through generated paintings.

In recent years, alternating current (AC) 50/60 Hz magnetic therapy has attracted attention for improving blood circulation and recovery from fatigue. However, its effectiveness in treating bedsores (pressure sores and ulcers) in bedridden patients has not yet been fully verified. In this study, we investigated the effect of an AC magnetic field (50 Hz, B_rms127 mT) on poor blood circulation induced in the rearfoot region of the sole near the heel of the left foot. Ankle weights (3 kg) were attached to the left lower leg, to reproduce the conditions for early stage of pressure injury. Skin blood flow volume in the sole was monitored for 70 min using 2D laser speckle flowmetry in six healthy adult males recruited for this study. Each participant was positioned supine on a bed, and the left lower leg was compressed and immobilized with ankle weights for 60 min, causing a decrease in skin blood flow volume. The heel and ankle were exposed to a magnetic field for 10 min, and the rate of change in the skin blood flow volume in the rearfoot region of the sole near the heel was analyzed. The results indicated that the magnetic field rapidly increased and restored the decreased blood flow volume, and a significant effect was sustained intermittently for 40 min after the end of exposure.

This study addresses the need for designing paediatric size powered prosthetic elbow. It evaluates an anti-parallel link mechanism driven by cables, aiming to minimize the force required for operation while ensuring a suitable Range of Motion (RoM). The mechanism, inspired by robotic arm designs, undergoes static and kinematic analyses to determine optimal design parameters. Results indicate that the mechanism can achieve a 180-degree flexion movement and is suitable for children, maintaining necessary force and motion capabilities. The optimized design parameters ensure the prosthetic's dimensions are appropriate for a 9-year-old child, making it a viable solution for paediatric use.

Wrist joint disorders significantly impair the quality of life (QOL) and activities of daily living (ADL). The primary treatments for such conditions typically involve partial or total wrist arthrodesis. While these methods are effective in providing pain relief and wrist stability, they result in the complete loss of wrist range. To address this issue, this study aims to contribute to the future development of artificial total wrists that closely mimic natural motion, with the goal of preventing declines in QOL and ADL. However, a major challenge lies in the lack of standardized metrics for wrist evaluation to guide the design of artificial total wrists. To tackle this issue, this research focuses on developing a simplified 3-D measurement method to establish benchmarks for comparing the motion of artificial total wrists with that of natural human wrists. Specifically, we construct a custom measurement device and reference tool, enabling the three-dimensional tracking of wrist motion using two cameras. The reference tool simulates the flexion motion of the right hand, and the measurement process involves recording the flexion motion of the right wrist. By employing programming-based analysis, marker positions are recorded at 0.1 second intervals. The measurement and analysis process is applied bilaterally and conducted with three participants, allowing for visualization of behavioral differences due to individual variability. The proposed method offers significant advantages, including its low cost—approximately 1/180 of the cost of conventional methods—while enabling visualization of motion trajectories. Despite its simplicity, the measurement accuracy is comparable to existing techniques, making it a promising approach for the development of artificial total wrists that can replicate the natural motion of the human wrist.

In this study, the application of electrical stimulation known to cause muscle cramps to simulate a gastrocnemius cramp during stair climbing was investigated. If gastrocnemius cramps can be consistently and repeatedly generated in this condition, an environment to simulate cramping while wearing a lower limb exoskeleton during a dangerous motion assist condition is possible. This type of environment can be used to develop exoskeleton control responses to involuntary muscle activity from cramp. To the author’s knowledge, there have been no studies done that have attempted to electrically elicit a medial gastrocnemius muscle cramp in any position other than lying prone. This study is the first attempt to apply electrical cramp induction to the gastrocnemius in multiple positions and motions that occur in stair climbing. These experimental conditions were used to confirm if cramp generation is still possible, consistent, and what the limitations of electrically elicited cramp are in this application. We find that electrical stimulation is ineffective to produce a gastrocnemius cramp during stair climb due to a limited time window to produce the cramp, a stimulation duration of multiple seconds needed to form a cramp, and a noticeable increase in resistance to cramp production from electrical induction while standing. However, this method could still be viable if additional conditions known to lead to a cramp prone state are additionally applied, or the subjects are all naturally susceptible to cramping.

In this study, we conducted an experiment to clarify the effects of different player arrangements used in bird's-eye view training (actual game layout and random layout) on the training effectiveness, with the aim of developing a training method to improve the cognitive ability to perceive a situation from a bird's-eye view in soccer. The results of the experiment suggested that the proposed training method using the actual game layout may be effective in improving the cognitive bird's-eye view ability of soccer players, enabling them to recognize the game situation faster and more accurately.

When a robot assists a human in a sit-to-stand motion, if the robot can unconsciously guide the human motion intention to the desired motion, it will be easier for the human-robot system to perform the desired motion in a coordinated manner. We focused on motor contagion due to prediction errors as a method of unconsciously guiding motion, and investigated its effect on the change in human motion during sit-to-stand motion by grasping the hand of a robot, which is similar to the robotic sit-to-stand motion assistance. After showing a video of a person standing up with the hand reaching diagonally downwards under the prediction that the person would stand up by reaching the hand horizontally, the standing trajectory of the experimental participants changed upwards. This result suggests that motion guidance through motor contagion is possible in robotic sit-to-stand assistance situations, despite the need to verify the effectiveness of this method by combining it with actual robotic assistive movements in the future.

A ring-type fingertip force sensor, focusing on changes in the tension of the flexor tendon, has been developed as a method to measure the force applied to the fingertip without obstructing fingertip sensation. In this study, to verify the effectiveness of the sensor on multiple subjects, an estimation method was investigated. This method combined two mechanical models – a sensor model showing the relationship between sensor output and the sensor protrusion, and a finger model showing the relationship between tendon tension and fingertip force with a neural network model analyzing the relationship between the reaction force on the protrusion and the tendon tension. A fingertip force estimation experiment was conducted on 10 subjects. As a result, while fingertip force could be estimated with a certain degree of accuracy, an unstable estimation phenomenon caused by the finger model was observed, indicating that further examination of this model is necessary in the future.

Fingertip dexterity is attracting attention in the fields of education and rehabilitation. Especially in writing operations used in daily life, the balance between the force applied from the fingertips and the reaction force from the writing instrument is important. However, conventional proficiency methods do not focus on fingertip force, and the relationship between grasping force and writing posture has not been fully elucidated. In this study, we aim to develop a learning system that focuses on fingertip force, and investigate the relationship between grasping force and writing posture. We will measure the writing direction and the grasping force at that time, and conduct experiments on the relationship between them.

This study evaluates the feasibility of an Environment-Adaptive Vibrotactile Transformation (EAVT) method, which identifies a tool’s transfer function and predicts vibrations for tools made of different materials or masses. We used acceleration data recorded with an ABS resin probe as the reference and attempted to estimate the recorded accelerations of heavier probes made of iron or equipped with a vibrotactile actuator. We assessed both the frequency spectra and subjective similarity of the predicted vibrations. The results indicate that while EAVT can partially account for amplitude changes due to additional mass, spectral mismatches remain problematic, leading to lower subjective similarity.

With the rapid expansion of the Internet advertising market, advertisements have become an indispensable part of modern society as a means of information communication. However, advertisements displayed during work are often considered annoying or disturbing, raising the concern that they may reduce the effectiveness of the advertisement. Personalized advertisements have attracted attention as a means of solving these problems, as they are expected to promote positive emotions and effectiveness through content based on individual interests and preferences. In this study, we conducted an experiment to investigate the effects of personalized advertisements in a passive work environment with different loads, to test the potential of advertisements as a work-aid tool. The effects of personalized advertisements on workers' emotions and work efficiency were evaluated using a text reading task and double-tasking. The results showed that in passive tasks with a light load, personalized advertisements promoted pleasant emotions and made workers feel less disturbed by advertisements despite decreasing their concentration. In heavy-load passive tasks, a relaxing effect was observed, suggesting that work performance was maintained. These results indicate that personalized advertisements contribute to the promotion of pleasant feelings and relaxation and may be used as a work-aid tool that exerts an emotional effect, especially in passive work environments.

To improve work efficiency and address the labor shortages caused by the declining birth rate in recent years, we developed a voice navigation system that enables a single person to perform maintenance and operation work on electric power facilities, which is typically performed by a team of two people. We conducted an evaluation experiment, in which the task was to shut down a high-voltage power-receiving facility. A comparison was made between the conventional method, in which two individuals worked in pairs, and the system condition, in which the voice navigation system developed in this study was used. The system conditions proved advantageous in that work could be performed without time pressure; stable performance was achieved regardless of the instructor; instruction misstatements could be eliminated; and time efficiency was better when a single person worked using the voice navigation system than two people working together. However, the disadvantages included an increased sense of burden on the worker and frustration caused by the long waiting times for the system to respond. We believe that the practicality of the voice navigation system can be improved by developing a system that supports the operator's perceptual activities and reduces the response waiting time.

Precise motion can be learned in many situations, such as dance, traditional crafts, and sports. The motion learning process can be divided into three stages, and this study focused on the recognition stage, which is the earliest stage of learning. In the recognition stage, motions are commonly learned by “imitation learning.” An imitation learning system that superimposes an instructor and a learner on a virtual reality (VR) platform was proposed in a previous study. In this study, we conducted a comparison experiment between first- and third-person views in VR imitation learning to evaluate their effects on VR imitation learning of full-body motions. The experimental results suggest that imitation learning in the order of third-, first-, and third-person views may be more efficient than imitation learning from a fixed view.

Crushed raw and intermediate materials in the cement production process were transferred from one place to the other inside the cement production plant. They occasionally fell from the conveyor and stacked on the ground to form solidified sediments in many places inside the plant. These solidified sediments should be removed before they started to interfere production facilities. We accordingly designed and developed a prototype of robotic crusher of hardened sediments. It consists of a three degrees-of-freedom manipulator mounted on a crawler type mobile robot. An electric breaker that crushes the sediments is attached at the end effector of the manipulator. Since large reaction force might be exerted to the manipulator during contact of the tip of the breaker to the sediments, we thought that compensation of reaction force is necessary for completion of crushing task. We accordingly employed a disturbance state observer on the control system of the manipulator and used the estimated disturbance for compensation. Both numerical simulations and the result of an experiment indicate the validity of our approach.

Motion assist robots have been studied to assist human motion for elderly, disabled, injured persons. It is important for the robots to estimate motion intention of users in real-time to assist their daily motion effectively. This paper presents a real-time method to estimate human lower-limb motion intention and its motion phase, enabling motion assist robots to determine when and how to assist human motion, even in the middle of the action. In the proposed method, integrated information about human physical and biological data and the surrounding environment are used to estimate the motion intention and its motion phase. Artificial Neural Networks (ANNs) with damping neuron has been applied to estimate the intended motion and its motion phase at the same time in real-time effectively. Two types of ANNs are used for motion intention estimation: a multilayer ANN and a recurrent ANN, and the estimation is performed using both results by switching them depending on the motion phase. Furthermore, changing velocity of the output from the artificial neural network is considered to figure out the intended motion in early stage. Experimental results show that lower-limb motion and its phase are effectively estimated with the proposed method. Furthermore, it is shown that the motion phase is correctly estimated with the proposed method even if the user gives up the intended motion during the motion.

Time-varying network communication delays cause significant challenges, such as reduced control performance and destabilization of closed-loop systems in networked control system. Addressing these issues is critical for achieving stable and efficient operation. A three-dimensional ball-on-plate system was developed using a six-degree-of-freedom industrial manipulator in this study . This system faces two main sources of delay: the round-trip time of network packets and the latency caused by observing the ball's position with a camera. These delays are time-varying and difficult to predict, potentially impairing the system's ability to maintain stability. a control approach was proposed to mitigate these effects. The time-varying delays were predicted using a Gated Recurrent Unit (GRU), a type of recurrent neural network known for its capability to model sequential data. The predicted delays were incorporated into a Smith predictor-based controller, which is well-suited for systems with known delays. This integrated control strategy was implemented to stabilize the ball-on-plate system. Experimental results demonstrated the feasibility and effectiveness of the proposed method compared to feedback control without delay compensation.

Robots equipped with measuring instruments are expected to automate inspections of infrastructure structures such as piping. In this study, a one-to-one robot simulation environment with a real-world environment was constructed in order to investigate a robot movement control method in curved and branching pipes, confirm the robot mechanism design, and acquire data for movement control. In this report, we describe the construction of a movement control algorithm based on the robot's position and joint angle information calculated by running the robot along a target trajectory in the simulator.

In recent years, worker shortages due to aging infrastructure structures and a declining population have become a serious problem. Regular inspections and repairs require inspection work at high altitudes and in narrow spaces, which calls for the use of robots. In this study, we aim to develop a negative pressure adhesion type wall-climbing robot, and are building an exhaust performance evaluation experiment system to be used in the design of negative pressure adhesion mechanisms. To improve measurement accuracy, we have made improvements to the airtightness of the experimental vessel to reduce leaks, which was an issue, by changing the vessel shape and reducing joints. Evaluation experiments confirmed the high airtightness of the vessel, and we will describe the results, which confirmed the usefulness of the exhaust performance evaluation experiment system.

As the use of 3D printing technology increases, various applications in engineering design are advancing. However, research on the influence of different shapes on the interests of engineering students is limited. This study aims to investigate the shape preferences of students from different engineering fields, including mechanical engineering, electrical engineering, and materials engineering, toward 3D printed models. A survey was conducted with 203 students from the National Institute of Technology, Tokyo College, asking about their preferences for four 3D printed models with different characteristics. The results show that shape features such as "soft," "smooth," "symmetrical," "transparent," and "round" were widely preferred across all fields. However, preferences for specific characteristics such as "cute" and "technical" varied by field. These findings suggest that designing 3D printed models with universal features can appeal to a broad audience, while considering field-specific preferences is crucial when targeting a particular group.

This study proposes a high-accuracy model for predicting human operations in complex tasks using a neural network (NN). By recursively incorporating task information, the NN predicts torque commands and simulates a plate control task (PCT) without relying on real-time human input. The PCT involves controlling a two-degree-of-freedom virtual plate with a rolling ball, driven by gravity and air resistance, while following various target trajectories. Experimental data were collected from ten participants performing three different target trajectories. Training sets were used to build models, which were then evaluated on separate datasets. Comparative analyses with a proportional-derivative (PD) model and McRuer’s crossover model showed that the proposed NN-based method achieved significantly lower prediction errors in torque, plate angle, and ball position. These findings indicate that this recursive NN model is well-suited for capturing complex human-in-the-loop behaviors and can enhance assistive technologies in advanced human-machine interaction systems. It operates quickly for real-time use.

To realize tactile VR, we are attempting to utilize Velvet Hand Illusion (VHI), so we investigated Gestalt factors including temperature condition. To progress this investigation, we developed a new VHI presentation equipment which possesses two thin brass tubes: one of two was passed cold water and the other was passed warm water. We performed a series of psychophysical experiments using the equipment according to semantic difference method (SD method) and method of limits. As the result of the series of experiments, it is found that when the two tubes were at different temperatures, smooth sensation was decreased by means of decrease of similarity factor, and that sticky and soft sensation were increased by increase of temperature.

It is known that vibratory stimulation to tendons can elicit various reflexes and illusions. For example, when vibratory stimulation is applied to the flexor tendon of the wrist, phenomena such as kinesthetic illusion (KI), where the wrist feels as if it moves in the extension direction without actual movement, antagonist vibratory response (AVR), where the wrist moves in the extension direction, tonic vibration reflex (TVR), where the wrist moves in the flexion direction, or no phenomenon at all, may occur. However, previous studies have not identified a method to selectively elicit these phenomena. This study investigates methods for selectively eliciting these phenomena.

In recent years，the demand for autonomous robots in the field of logistics and manufacturing has been increasing due to the shortage of manpower and the increasing demand for e-commerce．One of the solutions to the shortage of manpower is the automation of transportation using autonomous mobile robots. However, in the field, there are a wide variety of caster-equipped carts with different characteristics, such as approximate shape, caster installation positions, and combinations of swivel and fixed casters, so that caster-equipped carts are difficult to control position. In logistics and manufacturing, four-wheeled carts with casters are commonly used, typically featuring either four swivel casters or two swivel and two fixed casters. When transporting with an Autonomous Mobile Robot （AMR）, the position of fixed casters must be considered during rotation. Additionally, when moving caster-equipped carts, the overall dimensions must be taken into account to avoid collisions with surrounding objects. To address these challenges, we have developed a function that handle both four swivel casters and two swivel plus two fixed casters, ensuring obstacle avoidance tailored to each configuration. This development switches parameters such as rotation center, minimum turning radius, and minimum proximity distance to obstacles for each cart type, enabling collision-free transport and enhancing productivity.

Wearable sensors placed on the surface of human skin, capable of collecting and analyzing biological information such as body fluids, gases, and blood, are essential technologies for monitoring health status and assessing disease symptoms. Microfluidic chips, with their ability to rapidly collect, store, and analyze samples, demonstrate unique advantages among various health monitoring methods. However, due to the smooth and curved nature of human skin, achieving adaptable deployment and customized microfluidic chip designs has become a critical research challenge. To address this, we developed a rapid prototyping method for flexible microfluidic chips using DLP 3D printing technology. By employing biocompatible flexible photopolymer resin, we successfully fabricated microchannels with micron-scale precision. This low-cost chip demonstrates excellent adaptability to the curved surfaces of different individuals' skin and, when integrated with sensors, enables precise health monitoring. This approach provides an innovative solution to the complex manufacturing processes of microfluidic chips, paving the way for advancements in personalized medicine and wearable devices.

This is an activity report in 2024 of the industry-academia-government-finance collaborations for manufacturing DX that has been jointly planned and promoted by Musashino Bank and Toyo University and developed in Saitama area since 2018. As reported in IIP2023 and IIP2024, this academy is based on the combination of lectures, exercises, factory visit tour and discussions among participants. Participants from companies join in this group for solving the company’s own technical issues. Final goal of each participant companies is to improve productivity or to develop preventive maintenance by learning in this academy. In this report, I briefly report its activity in 2024 with some considerations on how the “space for the collaborative creation” are promoted for engineers in different companies.

A total of approximately 6,400 PowerPoint slides covering manufacturing technology, production management, data science, etc. were used in the "Production Systems Design" course at Tokyo Metropolitan College of Industrial Technology from 2020 to 2024. Student reports on these lecture slides were analyzed, and the following trends were found: 1) There is high interest in 3D printers, CNC, laser processing, and Gigacast. 2) 70% of students had experience using IoT, and 50% of students expressed a desire to be actively involved with it after employment. If negative attitudes are included, 90% of students were positive. 3) Regarding AI, there is a high level of interest in AI after employment. 55% of students expressed a desire to be actively involved, and if negative attitudes were included, 95% of students were positive.4) The social issues that generated the greatest interest were the energy crisis, global warming, global population trends, soil pollution, and marine plastics.5) The fields of application of IoT proposed by students were smart factories (35%), agriculture and forestry (20%), food and beverage/logistics (20%), and when medical and welfare are added, service work accounted for 35%, showing an expansion in the field of IoT applications.6) Video lecture materials were easy to understand (40%) or somewhat easy to understand (45%), so the use of explanatory videos of about 3 minutes was effective.

Currently, there is a strong need for developing a system for monitoring psychological stress monitoring system at work to improve the quality of life (QoL). This study proposes a system that records and analyzes work stress based on autonomic nervous system activity estimated from daily ECG measurements. The ECG signals are continuously recorded and transmitted to a server terminal, where they undergo preprocessing, including noise reduction to eliminate motion artifacts. The processed signals are then analyzed to extract heart rate variability (HRV) and compute the ratio of low-frequency (LF) to high-frequency (HF) HRV which is used as a stress indicator. The system was evaluated by measurements under stress stimulus and showed satisfactory performance. This real-time monitoring system enables continuous assessment of stress levels in a work environment.

Robots are required to automate the inspection of underground infrastructure structures. Therefore, a high-precision self-position estimation technology is needed that does not require line measurements and has a margin of error of only a few millimeters per several tens of meters. In a previous study, visual odometry (VO), which estimates self-position from captured video images, was investigated. The accumulated error between the camera mounting angle and VO processing was identified as an issue. The objective of this study was to increase accuracy by eliminating mounting angle errors and modifying VO processing. This paper reports the results of the study on the calibration function and the change of feature points used.