One of the issues in a super-aged society is the occurrence of wheelchair tipping accidents. While wheelchairs equipped with forward tipping prevention mechanisms have been proposed in the past, there are no wheelchairs that address lateral tipping. In this study, we analyzed the phenomenon of tipping during wheelchair use and proposed models for both forward and lateral tipping of wheelchairs. Subsequently, we developed a new anti-tipping caster that prevents both forward and lateral tipping of wheelchairs. Finally, through tipping experiments using multiple wheelchairs, we demonstrated the effectiveness of the proposed device.

The purpose of this study is to develop a new hands-free crutch made from paper. Hands-free crutches are useful for improving the quality of life of people with lower limb injuries, but they must be expensive when optimized to the body dimensions of the injured person. We therefore focused on paper as a recyclable and easy-to-process material. Then we conducted buckling load tests on a square pillar made from paperboard and a commercially available paper tube. As a result, the buckling load test of the paperboard quadrangle columns confirmed that the use of the manufactured quadrangle columns as an apparatus material was inappropriate due to the labor required to produce the quadrangle columns and the instability of their strength. However, we confirmed that commercially available paper tubes have sufficient strength. Since we were able to confirm the strength of the paper tubes through buckling load tests, we thought that by conducting bending load tests and strain measurement tests on the paper tubes, we could better guarantee the strength of the material and use the paper tubes as a material for the apparatus.

Sensory evaluation is often used for usability, and very few studies have focused on the mechanical load acting on handrails. In this study, we measured and analyzed gripping time and mechanical loads such as grip strength, tensile force, and torques acting on the stair's handrail during climbing up and down motion using a developed measurement system. Right-handed 4 healthy adults participated in the experiment by walking upstairs when using their right hand. As a result, the handrail gripping time was approximately 1.42±0.29 s, the grip strength ratio was approximately 23.7% of the maximum gripping, and the vertical downward load was -0.077±0.021 N/N during climbing down. Also, the handrail gripping time was approximately 1.43±0.34 s, the grip strength ratio was approximately 23.7% of the maximum gripping, and the vertical downward load was -0.074±0.035 N/N during climbing up. In future studies, we plan to motion analyze using the mechanical load impulse and body joint angles.

To apply the three-axis tactile sensor to various objectives, we developed the sensor which mimicked human index finger of almost similar size of human one. The sensor finger pad is an acrylic core covered with a thin silicone rubber sheet; normal and tangential forces applied to the finger pad are obtained from gray scale value and its centroid movement observed with a miniature CMOS camera. To estimate effectiveness of this sensor, the sensor is mounted on a human hand shaped gripper and the gripper grips a cup stepwise elevating its weight with input of 10-yen coins one by one. Consequently, we conclude that this sensor possesses sensing capabilities of normal and tangential forces because the gripper can enhance its grasping force according to weight elevation of the cup.

We develop a tactile sensor that estimates the three-axis force applied to the finger pad from the nail color, utilizing the relationship between the deformation of the finger pad and the change in nail color. To improve the accuracy of this sensor, it is important to develop experimental equipment that can more precisely measure the three-axis force applied to the finger pad. The experimental equipment used to create datasets in previous studies restrained one of the three axes and only measured two-axis forces, which meant that the force applied in the restrained direction was unknown. In this study, we develop a measuring apparatus that can simultaneously measure three-axis forces, collect a new dataset, and attempt to improve the accuracy of the nail color sensor.

Various nursing care motions are conducted by nursing care givers. Some of them should be done with proper postures, because doing such nurcing care motions with improper posture can cause physical burdens for care givers. For example, a caregiver is required to crouch when the caregiver assists a trancefer motion of a patient. If the caregiver does not croutch enough, it is expected that the caregiver is notified about the improper posture for correcting the improper posture; the caregiver can know its own improper posture, and it may prevent the physical burden. In this study, an estimation method of a caregiver’s back height from the plantar pressure distributions was proposed. Experiments were conducted for evaluation of the proposed method.

Japan has been actively working to improve the earthquake resistance of homes and buildings after multiple major earthquakes have occurred in the past, resulting in the collapse of buildings. While research has been conducted on inspection robots for spaces such as under the first floor of houses and under the ceilings of buildings, inspection robots have not been developed for narrow spaces such as second-floor attics because of the limited range of human observation. This research aims to develop an automatic percussion mechanism with an acoustic analysis function that can be equipped with a residential inspection robot for use in narrow spaces.

The safety and reliability of the vascular access puncture are not only fundamental in hemodialysis treatment but also critical for reducing the patients' burden. Proper insertion depth sensing is demanded to improve the quality of the needle placement in the hemodialysis puncture. This manuscript introduces an advanced approach to measuring the needle insertion depth utilizing a laser-ranging device and an inertial sensor mounted on the puncture needle. In the initial stage of the research, the prototype of the measurement equipment was demonstrated to validate the theoretical correctness and availability under the hand-manipulating punctures. The proposed method successfully assesses insertion depth by correcting the deviations due to three-dimensional needle posture and sensor mount alignment errors.

Gordon's functional health pattern-based assessment, one of the frameworks of nursing diagnosis, is used to identify health problems and risks in patients at an early stage by repeatedly conducting systematic data collection, analysis, professional judgment, and estimation from 11 different perspectives. However, although this nursing diagnosis is used as an educational tool and manual to understand the essence of nursing, little progress has been made in rationalizing and improving efficiency through computerization and automation. In this paper, we introduce a method of using AI and robotics technology to assist in the automation of assessment in nursing diagnosis.

In this study, we use acceleration sensors to compare the characteristics of normal gait and abnormal gait such as circumduction gait. The acceleration sensor output includes translational, centrifugal, tangential, and gravitational accelerations. When the body part to which the sensor is attached changes its posture significantly, the gravitational acceleration is included in the output of each axis of the sensor. It is difficult to evaluate the acceleration caused only by walking motion. Therefore, by sequentially estimating changes in the posture of the sensor, gravitational acceleration is removed from the sensor output, translational, centrifugal, and tangential accelerations which are caused only by walking motion are evaluated. Furthermore, performing singular value decomposition on the translational, centrifugal, and tangential accelerations and evaluating the coordination of accelerations are conducted to distinguish normal and abnormal gait.

It is suggested that ground reaction force be employed as a useful quantitative parameter for the evaluation of the effects of exercise on self-monitoring. The objective of this study is to develop an effective method for estimating ground reaction force with accuracy and reliability. In a previous report, it was proposed that a method for estimating three directions of ground reaction force from measured gait image could be developed using convolutional neural networks (CNN), a type of deep learning. The results showed that the proposed method is an effective approach for estimating the three-dimensional ground reaction force with high accuracy in a fixed-point camera. The objective of this paper is to estimate the vertical ground reaction force with variable camera positions. Computer graphics (CG) were created from motion data obtained during the gait process. Training data was captured in CG from different camera positions and used to develop a deep learning model. The estimation accuracy was then verified. As a result, there were camera positions that could be estimated with some degree of accuracy. It was shown that ground reaction force estimation with variable camera positions during estimation is feasible by using CG.

In this study, we construct a model describing the relationship between gait velocity and muscle forces of lower limbs, which would discriminate between normal and abnormal gait. Walking is a movement that is based on inter-articular coordination of the lower limbs. In maintaining and improving walking ability, it is important not only to maintain individual muscle mass, but also to use muscles that enable proper inter-articular coordination. In people with leg disabilities and the elderly, gait is different from normal gait by a healthy person. In the measurement experiment, normal, scissors and circumduction gaits performed by two healthy adult males have been measured using an optical motion analyzer. The two individuals walk three times each, for a total of nine times. The gait model is constructed using some muscles of the left and right lower limbs. The application of the gait model to discriminate between normal and abnormal gait is examined based on the accuracy of the estimation of center-of-gravity velocity during walking. Gait models that have been constructed for each gait show potential for identifying abnormal walking.

In this paper, assistance method when using a standing assistance device is investigated. This device is intended for elderly people who have difficulty standing up, and the chair seat is designed to trace a trajectory that reduces joint moment. Since the standing up motion differs from person to person, it is necessary to assist each person at an optimal speed. However, it is difficult to estimate the optimal speed by having elderly people try standing up motion multiple times. Therefore, we proposed a new simulation method of standing up motion. We also investigated the effectiveness of simulation model by comparing it with the former studied model and actual standing up experiments.

We aim to create an environment for evaluating walking assistive devices under conditions similar to those experienced by actual users, using bipedal robots that simulate the gait of elderly individuals. We focused on hallux valgus, a common cause of abnormal gait, and developed a robot foot that replicates both the shape of hallux valgus and the characteristics of the foot arch. By measuring plantar pressure with pressure sensors, we assessed whether the arch function of the robot foot accurately mimics that of a foot affected by hallux valgus. As a result, the robot foot reduced plantar pressure on the big toe and increased pressure on the heel and lateral side. This indicates that the robot foot successfully replicates the arch function of a foot affected by hallux valgus.

Apple harvesting is the most prosperous agriculture in Aomori prefecture. However, the labor shortages and the increase of deserted arable lands due to lack of the new producers are the issues in recent years. To prepare a healthy and safe working environment for producers, workload should be measured to assess their fatigue and safety. As a way to investigate the characteristics of physical activity, surface Electro Myography (sEMG) is one of the methods used to measuring muscle activity. However, this method is not suitable for long-term measurement due to noise generation caused by perspiration, and it needs some advance preparation such as hair removal as also. Force Myography (FMG), which detects changes in muscle thickness and elevation of the skin, has been proposed as a method for evaluating muscle activity. Also in the previous study, a “muscle stiffness sensor” was developed to estimate the change in muscle stiffness during motion. In this study, we created an experimental system model which simulate a fruit tree joint system considered as one of the less-workload tree systems, and then evaluate the physical load in harvesting; one of the cultivation process, by using muscle stiffness sensors. The experiments were conducted at 1m and 1.5m height stems. In this experiment, it was found that the physical load during apple harvesting in 1.5 m height stem could be lower than in 1.0 m height stem, by using effective use of stepladder.

We are developing a Human-Machine Interface using surface Electromyogram (sEMG) for hemiplegic patients. Our approach employs machine learning to estimate the direction of shoulder movements from sEMG signals directed toward a target. In this paper, we aim to improve accuracy by using 35 features and employing a feature selection algorithm to identify highly contributive features. As a result, the accuracy with the initial 35 features was 62%. By utilizing a feature selection algorithm to select features with high contribution rates, the accuracy improved to approximately 76%.

Human Machine Interfaces (HMIs) have seen significant advancements recently, especially in aiding communication for patients with amyotrophic lateral sclerosis (ALS) and quadriplegia due to nerve damage. Among these, HMIs utilizing tongue movements have been proposed as a method leveraging relatively intact body parts. Specifically, weak electromyogram (EMG) signals during tongue motion serve as an effective input interface due to their higher signal-to-noise ratio compared to brain waves, allowing for stable measurements and reducing user burden. However, estimating consonants from EMG signals remains challenging. This study aims to develop a speech recognition system enabling non-vocal communication, utilizing EMG signals measured around the hyoid bone and leveraging advanced deep learning techniques. The proposed system integrates Convolutional Neural Networks (CNNs) for vowel estimation and Long Short-Term Memory (LSTM) networks for word prediction. The experimental methods include a detailed system overview, explanations of CNN and LSTM architectures, and a comparison of label estimation accuracy using different CNN parameters. Additionally, sequence data prediction using LSTM models is described. The experimental results demonstrate that combining the vowel estimation CNN with the word estimation LSTM yields a highly generalized model, enabling efficient and accurate non-vocal communication for patients. This HMI software provides a promising solution for enhancing patient communication capabilities.

In this study, we prototyped the system that collects data of the movement related to falling for the analysis of the elderly persons’ falls. Focusing on the head height and acceleration of front waist as indicators for analysis of falls, we connected sensors that collects these data and the system. After detecting falls by using fall detection algorithms based on the collected data, the system records data obtained from sensors and creates time series data, metadata, and traffic flow data as data for fall analysis. As an evaluation of the system's performance in collecting data on falls, we checked fall detection accuracy rate and the operation of the system when performing simulated falls and motions that are likely to be considered falls. As the result of the experiment, in three types of sensor systems, the system correctly detected simulated fallings 83 % or more and created data for analysis. In this paper, we confirmed the system could collect the movement progress data required for analysis of falls.

In this paper, we propose a method for controlling following vehicles based on image processing and machine learning using an object detection algorithm (FOMO). In this study, we develop a cooperative work algorithm that combines image processing and communication using a small robot equipped with a microcomputer and a small camera that can communicate wirelessly. First, we describe a method for estimating the distance to the leading vehicle and the attitude angle of the following vehicle by machine learning of logo images using Edge Impulse. Next, we explain a simple PID control law that can maintain the distance and attitude angle between the vehicles. The proposed method is characterized by sharing motor speed commands via wireless communication, and we propose a control algorithm that enables stable following by combining the sharing method with the control law for distance and attitude. Finally, we evaluate the tracking accuracy by acquiring the trajectory of an actual driving experiment.

In this study, we propose a static or dynamic obstacle detection and avoidance method for a person-following robot. We developed a person-following robot with three thermal array sensors and four omniwheels. The human-following robot is equipped with 2DLiDAR to obtain point cloud data of obstacles in the vicinity. If a certain number of points in the obtained point cloud data are in the obstacle detection area, the robot recognizes the obstacle and generates an avoidance maneuver. Finally, the static and dynamic obstacle detection and avoidance performance were verified, and an obstacle avoidance method with high obstacle avoidance performance was successfully devised.

Tonotopy plays an important role in sound perception. It functions to activate specific areas of the primary auditory cortex at different frequencies. This study supposes that the brain is activated differently in the primary auditory cortex even for sounds with frequency difference so small that human cannot discriminate them in consciousness. The objective is to use fMRI to capture brain images by sound stimuli with small frequency difference, and to use deep learning to identify the sounds that human hear. In a previous report used the block design and the event-related design for the imaging design. The results showed that the block design produced a high discrimination rate for two sounds with 5 Hz difference. However, the discrimination rate for the two sounds with 1 Hz difference was low for both designs. The purpose of this report is to build a method that can discriminate two sounds with 1 Hz difference, which is low result. As the first step, experiment with increasing the number of training data in the block design. As the second step, it is generally said that the estimation rate improves as the number of training data increases in deep learning. Therefore, as a trial used a combination of both designs in the training phase and each design in the evaluation phase. From the results of a previous study and this study, it was found that the block design is suitable for discriminating two sounds with small frequency difference using deep learning.

In this paper, we proposed a system that can record a series of the fetal heart rate and the Bishop score, which related to the progress of the delivery, and display the necessary information based on the role of the persons who try to check the information during delivery. The system allows the nurses and doctors related to the delivery to check the progress at their own devices. We implemented the two elemental functions of the proposed system. One is an input function for recording delivery indicators eliminated device dependency of the input data, and the other is an output function for displaying the delivery information to the persons who require it. As a method for eliminating the device dependency, we used an API (Application Programming Interface) to acquire data. This allows different devices to be treated as a single system. The prototype system enables to input the necessary indicators to calculate the Bishop score in a quick and simple operation of 5.4 seconds. Moreover, the system can be used to check information without visiting the delivery room. Thus, the proposed system may reduce the time to record and check information.

This paper describes about clarification of auscultation sounds to improve a performance of a multipoint automatic auscultation system.During automatic auscultation, heart sounds and environmental sounds may be mixed in with lung sounds of interest, which can reduce accuracy of normal/abnormal discrimination. Taking advantage of the system's ability to simultaneously listen to sounds at multiple points on patient’s chest surface, a method to restore a characteristics of an original lung sounds from sounds that are a mixture of heart and lung sounds is proposed. This method is implemented in the system and verified its effectiveness.

The ring-type sensor can measure tendon tension by pressing a protrusion into the finger and can estimate the fingertip force without interfering with fingertip sensation. The purpose of this study was to establish a design procedure for the ring-type sensor by examining the suitable pressing depth of sensor into the finger. In the experiment, for 6 subjects, we performed the indentation tests on the finger with applying fingertip force of 10 N and measured pressing depths and pressing forces on the fingers that the subjects felt uncomfortable. As the results, it was found that the use of the curved part of the sensor reduced the subject's discomfort. From the result, we established the design procedure to determine the suitable pressing depth of curved part and protrusion of sensor.

The latest information and communication technologies (ICT) and artificial intelligence (AI) are beginning to be used in tasks to “confirm safety” that were previously performed by humans. The Digital Agency is, for example, reviewing “analog” regulations that require human involvement. Instead of using ICT or AI to support humans, ICT or AI is replacing humans. Therefore, the task of confirming safety that was performed by humans is expected to eventually be automated with ICT or AI. Given the history of automation and improved safety of human work by machinery and equipment, the introduction of ICT or AI in tasks to confirm safety will probably require designs based on risk assessment. The dangerous failures that may occur when ICT or AI performs tasks such as gathering information, making decisions, and transmitting information in place of humans need to be envisioned. Thus, the current work has attempted to use the safety confirmation type as a framework to identify safety design issues from a bird’s eye perspective.

The present study conducted a survey targeting both provider companies and users to evaluate effect of safety education using Virtual Reality (VR). VR is focusing as a useful tool to replace of traditional safety education methods of simulating dangerous situations without risk. However, quantitative evaluations of effective VR-based safety education have not been sufficiently conducted. In the present research, VR education on workers' safety behavior and safety awarenesses were surveyed in terms of both providers' and users' perspectives. The results revealed that the providers showed highly value of effectiveness. On the other hand, users did not always think its clear impact. The results indicate that the current state and challenges of VR safety education, and suggest that need consensus of provider and user for more effective VR-based educational programs.

Powered exoskeletons are expected to provide a means of independent walking for patients with spinal cord injuries (SCI). However, safety measures for physical therapists who assist with walking training in medical settings are insufficient. Safety measures are necessary for walking training care, which places a large physical and mental burden. In this study, we analyzed the walking training assistance actions of SCI patients wearing an exoskeleton and performed a risk assessment (RA). As a result, the number of unacceptable risks was reduced after RA compared to before RA. In particular, the number of risks in step 3 was greatly reduced, suggesting that RA is also effective in walking support settings.

For the safe design and manufacture of collaborative robots，there are standards such as the “Functional Safety Utilization Practice Manual”，ISO 12100，ISO 13849-1，2，etc.，but after introduction，safe collaborative work on site is required. In the present study，we conducted a preliminary study to set experimental conditions at a collaborative workplace where humans place parts and robots assemble the parts. The results suggested that the incidence of unsafe behavior may vary depending on the robot's assembly failure rate and the human work experience. In the future，we plan to conduct further studies advanced the present experiment，identify factors that arouse unsafe behavior in people，and consider safety measures.

In this study, we used a headlamp tester based on the inspection standards of the Road Transport Vehicle Act to investigate and analyze the headlamp irradiation characteristics of different light sources, HIR2 halogen lamps and bulb-type LEDs, and investigated and analyzed the correlation and irradiation characteristics of high beam and low beam.The experimental results showed that when switching to high beam, the bulb-type LEDs meet the inspection standards for low beam characteristics that meet the inspection standards, but the halogen lamps are highly likely not to meet them.The low beam irradiation luminance against the irradiation time decreased by about 5% after 120 minutes of irradiation for the HIR2 halogen lamps and about 15% for the bulb-type LEDs. The low beam irradiation direction against the irradiation time changed by 5 cm downward and 40 cm to the right for the halogen lamps, and by 2.5 cm downward for the bulb-type LEDs. There was no change in the irradiation direction in high beam for both immediately after turning on the lamps until 120 minutes later.

The accuracy of camera calibration by using combined with point and straight line method was compared with that of the point based method by performing camera calibration based on a pinhole camera model in the world coordinate system. Furthermore, the accuracy of the camera calibration by using the combined method was statistically evaluated in terms of the RMSE and the camera viewpoint using the simulation method described in the previous report. In camera calibration based of a pinhole camera model in the world coordinate system, the reproducibility of the camera viewpoint was good for both the point based method and combined method. And also, both methods had small variations in camera viewpoint relative to the RMSE. Even when the accuracy of the camera viewpoint relative to the RMSE in the point based simulation was poor, the combined simulation was able to reduce the error of the camera viewpoint relative to the RMSE.

When main line roads are traffic jam, many vehicles move into residential areal roads to avoid traffic jam. Furthermore, it is expected that safety will decrease due to the large number of vehicles passing on narrow residential areal roads. As a method to evaluate this, this study focused on the relationship between the occurrence of sudden braking and the vehicle traffic volume. Similar to previous studies, the road was divided into small sections of several meters each, and the relationship between sudden braking frequency and time zone, road section were analyzed. This analysis confirmed that there are cases in which sudden braking occurs frequently and cases in which sudden braking occurs less frequently during time zone when vehicle traffic volume on residential areal roads increases.

The concept of the vehicle has changed with technological innovations on last decade. Today we can call these changes basically as "CASE" (Connected, Autonomous/Automated, Shared, and Electric)．On the other hand, labor shortages are becoming more serious due to a decline in the domestic working population. To solve this problem, we developed "Telemotion" that is a remote-controlled autonomous vehicle driving system that automates the movement of vehicles made in factories. We have enabled safe unmanned autonomous driving and introduced it in our own factories by utilizing advanced technologies such as AI. We are currently building a digital twin environment that reproduces reality in a virtual space, with the aim of expanding this to other factories, and are proceeding with AI learning and system design in the digital. We will also show that development using this digital twin will solve problems related to labor shortages and productivity.

Airless tire (Non-pneumatic tire, NPT), which can achieve performance that cannot be achieved with pneumatic tires, are expected as tires that do not puncture. Airless tires with a Linked zigzag structure that has been proposed so far have the characteristic of being able to support the wheel load with the entire tire, and are expected to reduce rolling resistance. On the other hand, airless tires with a honeycomb structure have also been proposed. In this study, we conducted a mechanical analysis of both structures of airless tires, based on deformation modes and stress distribution, to examine the mechanism by which both structures support the wheel load and demonstrate the effectiveness of the proposed structures.

In recent years, research and development of autonomous parking systems has been conducted. The purpose of this paper is to develop an autonomous vehicle robot that can be applied to an autonomous parking system, which performs path following using RTK-GNSS (Real Time Kinematic - Global Navigation Satellite System) and obstacle avoidance using deep learning. RTK-GNSS was adopted because a highly accurate positioning system is necessary for path following. In addition, an object detection model based on YOLOv4-tiny deep neural network was employed to avoid obstacles on the path. First, the usefulness of RTK-GNSS was verified by comparing the positioning error between RTK-GNSS and stand-alone positioning. Next, the object detection model was trained to detect red traffic cones used as obstacles. Finally, an autonomous driving system for a vehicle robot was developed using RTK-GNSS and the object detection model. Then, autonomous driving experiments including obstacle avoidance were conducted to verify the usefulness of the autonomous driving system.

The present study examines the effectiveness of a driver assistance system to indicate the lane change possible area on the road by comparing a conventional assistance system that indicates information on the door mirror. The proposed and the conventional assistance systems are constructed on a driving simulator and evaluated through the simulator experiments. The proposed assistance system prevents dangerous lane changes when the intervehicle distance is relatively short, and encourages lane changes when the intervehicle distance relatively long. The proposed assistance system makes it easier for drivers to make lane changing decisions, reducing the number of lane change passes. In addition, the proposed assistance system shortens the time it takes to start a lane change, as the driver makes it possible to determine in advance whether a lane change is possible. As a result, the proposed assistance system enables safe and efficient lane changing.

In recent years, autonomous driving technologies in automobiles have been actively developing. When the operation of an autonomous vehicle differs from the preferences of its occupants, there is a possibility of stress occurring, particularly with braking operations, which significantly affect ride comfort. This stress amplifies negative psychological effects such as unpleasantness and anxiety, thereby reducing the inherent value of the vehicle. Due to this, it is meaningful to examine the preferred braking patterns of passengers to understand what constitutes comfortable braking in autonomous vehicles. Therefore, this study aims to quantify the sensory evaluation of braking by automobile occupants and to classify their preferences using the Semantic Differential method questionnaire survey. Additionally, we will focus on the occupants' posture maintenance movements and examine measurement methods for body behavior during actual vehicle operation. This report proposes an experimental approach to simultaneously measure sensory evaluation and body behavior, aiming to elucidate their relationship during braking.

In recent years，personal mobility devices such as electric scooters have been attracting attention as a means to solve environmental and transportation problems．In Japan，electric scooters have been gaining popularity through sharing services since the deregulation of the Road Traffic Law that came into effect in July 2023．However，there are concerns about the safety of electric scooters，and fatal accidents have been confirmed in Japan，with the number of accidents increasing every year．Single vehicle accidents account for a high percentage of all accidents，and tripping over steps is one of the causes．Because electric scooters have small wheels，they are more susceptible to steps，which can easily lead to accidents．As a measure，it is thought that accidents can be reduced by providing safe driving assist that makes drivers aware of hazards in advance．Therefore，in this study，considering the driving environment of the electric scooters，To use force-sensing stimuli that can be used as a means of danger perception even when the driver's vision is occupied with driving tasks．The force stimuli include handle vibration generated by a vibration motor．Simulation experiments are conducted to evaluate the effectiveness and timing of safe driving assist using haptic guidance. In this experiment，the assist was 25 meters and was the safest and highly evaluated．

Railway operators control track by inspecting track irregularity, which are the amount of deviation of the track from its original shape. As track is deformed by vehicle loads, dynamic irregularity inspection under vehicle load is desirable. However, some railway operators inspect only static irregularity. We study dynamic management methods for tracks using similar easily measurable in-service vehicle car-body. In this study, we take up dynamic irregularity, static irregularity, and car-body acceleration on the same track. We derive the ratio of dynamic irregularity to static irregularity, and after that, we investigate the relationship between the ratio and car-body acceleration. As a result, the tendency that the larger the ratio, the greater the car-body acceleration is observed.

This paper shows a case study of vibration measurement during the transportation of cultural properties from Tokyo to Naha. The author attempted to measure the vibration of the entire transportation process, from departure to arrival, using acceleration data loggers. Although the amount of data obtained was enormous, a semi-automated scheme was developed to calculate the raw data, and the time-series plots of acceleration and vibration frequency output were analyzed mainly in terms of trends in the occurrence of vibration frequency. This paper will serve as an example of an effort to look at time-series changes in vibration frequency from a bird's eye view throughout the entire transportation process.

As a method to mitigate the explosive increase in workload caused by the increasing complexity of electronic control systems, we propose a data-driven control with an adaptive mechanism. In this paper, we propose a two-degree-of-freedom control design that applies a machine learning model to the FF control unit, and describe the method of FF control calculation using the machine learning model.

Traffic flow simulation generally focuses on traffic phenomena such as traffic capacity optimization and congestion. Few studies have considered harmonization and interference with other vehicles in the vicinity of the vehicle. We have developed algorithms for steering and acceleration/deceleration control of autonomous vehicles using a risk potential model. In this paper, we apply this algorithm to multiple vsehicles and simulate a typical urban driving situation in which vehicles interact with each other to form a traffic flow, combining following driving and lane changing. The driver's driving behavior is determined by the risk potentials perceived from the preceding and surrounding vehicles. Since vehicle movement affects the driving behavior of other vehicles, a vehicle dynamics model was used to calculate the vehicle motion and trajectory. The results suggest that it is possible to simulate traffic flow that reflects the driver's driving behavior and to quantify the stress that the driver perceives from other vehicles.

This paper examines the fault tolerant control problem of a quadcopter with failed sensors. In the case where all state variables of control system are not exactly known, the control system needs to incorporate some type of state estimation. This paper applies the state estimation method based on Unscented Kalman Filter (UKF) to the observer system of a quadcopter with limited output sensors. Model predictive control (MPC) is a well-established control method in which the current control input is obtained by solving a finite-horizon open-loop optimal control problem using the current state of the system as the initial state, and this procedure is repeated at each sampling instant. The objective of this study is to establish a control method for a quadcopter with limited measurable state variables by means of incorporating state estimation into model predictive control.

This study aims to develop an autonomous driving system capable of navigating narrow L-shaped roads using deep reinforcement learning. While prior methods, such as the one proposed by Tahira et al. (2022), have demonstrated success in geometric path generation for narrow roads, they were limited by the constraint of constant maximum steering angles, resulting in some road widths being impassable. Our approach leverages flexible steering angle control through deep reinforcement learning, allowing the vehicle to learn optimal actions via trial and error. The trained model successfully navigated various narrow L-shaped roads, employing reversing maneuvers when necessary. In tests, the model achieved a 50.6% success rate compared to the physical limit trajectory in navigating right-angled L-shaped roads with entry and exit widths ranging from 2.0 to 5.0 meters. Although the success rate is lower than that of previous methods, this study demonstrates the potential of using deep reinforcement learning for more flexible and adaptable navigation.

Since the semi-trailer has a hitch point, the tractor pushes the trailer from the rear when backing up in the coupled state. And when the semi-trailer needs to turn, it is necessary to turn the steering wheel in the opposite direction of the desired turning direction. Therefore, the operation is extremely difficult and requires more complicated processing than that of an ordinary heavy-duty vehicle. Semitrailers, which require complex operations, have a high affinity with driver assistance systems such as automated parking systems. In this study, the reinforcement learning is applied to automated reverse parking of a semi-trailer. Reinforcement learning does not require training data and has the advantage of automatically finding the optimal action. A model of a semi-trailer is created in a simulation environment, and investigated whether the semi-trailer can acquire the ability to park backward using reinforcement learning. Sensors are placed at the center and four corners of the trailer and tractor, and rewards are set depending on whether the sensors are within the parking space or not. As a result, an automated reverse parking model of a semi-trailer was successfully built in the simulation.

Manually operated drones equipped with GPS (Global Positioning System) have problems such as difficulty avoiding obstacles in the absence of a GPS environment．On the other hand，drones equipped without GPS，which use SLAM (Simultaneous Localization and Mapping) for positioning control，have the problem of being too large．Therefore，the purpose of this research was to develop a method of flying in small spaces and to observe the environment with a small-sized drone using relatively simple image processing and autonomous flight without GPS．The base drone is a Tello EDU．The control program is an image processing program based on OpenCV (Open Source Computer Vision Library)．As a result，we succeeded in developing a drone that can recognize only specified colors such as green and yellow，and can observe the environment without collisions．The error between the area percentage of yellow and the theoretical value was within 25%．

This study examines the control problem of an underactuated ship. Model predictive control (MPC) is a type of optimal feedback control in which the control performance over a finite future is optimized with a performance index that has a moving initial time and terminal time. MPC enables optimization of control performance while taking the constraints on state and control variables into account. The objective of this study is to propose nonlinear MPC method for trajectory tracking and obstacle avoidance of an underactuated ship. The effectiveness of the proposed method is verified by numerical simulations.

Aviation gas turbine engines are required to be more efficient, due to consideration of environmental problems. The temperature of the combustor needs to be increased to achieve higher efficiency, and it is essential to improve the cooling performance of the high-pressure turbine. Film cooling plays a central role in high-pressure turbine cooling, and various film cooling hole configurations have been proposed. In this study, a new proposal is made to introduce a flow division promoter (FDP) into the crater hole. In the experiments, a heated wind tunnel apparatus was used at Ashikaga University. The experimental result was a bifurcation of the cooling air due to the FDP, which affected the counter rotating vortex pair (CRVP) and improved cooling performance in terms of area-averaged cooling efficiency and spanwise cooling efficiency.

To realize the goal of space transportation, that is a single-stage-to-orbit vehicle, not only technologies to operate the space transportation system frequently but also ones to innovate spaceflight performance are needed. Utilization of air-breathing engines in space transportation could innovate main propulsion systems of the space transportation system. We consider making a suborbital sounding rocket reusable with using the air-breathing engine, which results in realizing reusable sounding rocket in more compact scale. Furthermore, the realization of the reusable sounding rockets can also demonstrate the effectiveness of the air-breathing engines in space transportation. In this paper, system sizing and flight trajectory analysis of the reusable sounding rocket with the air breathing engines were conducted. From the analysis of the results, a relationship between the vehicle size and the flight performance was found. And also, the amount of propellant required for landing after deceleration by aerodynamic force was determined.

The authors are involved in the research and development of small hybrid rockets. An important component of launch experiments is the electronic system, which acquires experimental data on the rocket vehicle in flight, determines its position, and deploys a parachute during descent. In this study, a launch experiment with the 5^th small hybrid rocket was conducted in February 2024. During this experiment, after the rocket vehicle passed its apex, the apex was detected based on the acquired air pressure values, the parachute was deployed, the rocket vehicle decelerated, and it was recovered from the sea surface. In all four previous launch experiments, low communication frequency has been cited as an issue. Wireless communication is a reliable data acquisition method that does not depend on whether the rocket vehicle can be recovered or not. Since parachute deployment is now uncertain, recovering the data stored inside the aircraft is difficult, making the improvement of communication frequency the most important mission. Therefore, in this experiment, with the aim of improving communication frequency, a communication test was conducted by changing the method of acquiring position information and adjusting the settings of the communication module, which was then verified in the 5^th launch experiment

Hybrid rockets use different phases of oxidizer and fuel as propellants to generate thrust. Hybrid rockets have advantages over conventional rockets, such as higher safety, lower cost, and the ability to restart. However, they have the disadvantage of lower thrust compared to conventional type rockets. The authors conducted static fire test, using different fuel axial lengths and fuel geometries. For the fuel, poly methyl methacrylate (PMMA) is used in the former comparison, while PMMA and paraffin wax (PW) are used in the latter comparison. The experimental results show that the shorter fuel has better thrust characteristics in the comparison of fuel axial lengths. This will be explained by the increased length of the aft-combustion chamber, which promotes combustion. In the comparison of fuel geometry show that the hexagonal radial-hole fuel has better thrust characteristics. This will be explained by the larger time-averaged mass flow rate of fuel, it is about 4 times larger for the hexagonal radial-hole fuel than the single port fuel.