Keywords: Human-Robot Cooperation/Collaboration, Control Theory and Technology, Human-Robot/System Interaction
Abstract: As the application fields of Autonomous Mobile Robots (AMRs) expand, stress-free interaction with people around them becomes more important. AMRs are highly required to interact with pedestrians like human-human interaction. This paper proposes a new measure, 'Inducibility,' expressing how AMR's behavior can induce (control) the other pedestrian's behavior. In particular, this paper discusses two kinds of inducibility measures: one is based on the sensitivity of decision-making, and the other one is based on the controllability gramian for the state space model of the interactive behavior. Finally, the validity of these two measures is evaluated based on data on interaction behaviors, and typical applications of these measures are discussed.

Keywords: Modeling and Simulating Humans
Abstract: The development of robotic exoskeletons for human motion assistance and rehabilitation has intensified the need for efficient methods to discern users' motion intentions, particularly for complex upper limb movements. The shoulder joint poses the most significant challenge due to its anatomical complexity, multi-degree-of-freedom (DoF) movements, and variable muscle function. Consequently, real-time estimation of shoulder joint motion intentions remains the least explored. Existing prediction methods often fail to address the shoulder’s range of motion, as many models are restricted to below the shoulder level or rely on computationally intensive physiological parameters unsuitable for real-time applications. The shoulder's varying moment arm further complicates signal normalization and accuracy. To overcome these challenges, we propose a novel approach combining a Neuromusculoskeletal Model with a Multilayer Perceptron (MLP). This method leverages biological insights from the model to improve neural network performance, focusing exclusively on regression for real-time predictions. The method inputs include sEMG signals from shoulder muscles and shoulder angle data. These sEMG signals are preprocessed and integrated into the Neuromusculoskeletal Model, which incorporates activation dynamics, contraction dynamics, and musculoskeletal measurements. By correlating skeletal measurements with humeral elevation, the method reduces computational complexity and calculates muscle torques, which are then used by the MLP to estimate humeral acceleration. This approach effectively addresses normalization issues by correlating shoulder angle, sEMG data, and estimated torques with humeral acceleration.

Keywords: Sensor Fusion, Haptics and tactile sensors, Medical Systems
Abstract: This paper contributes to realize the surgical training simulator for the chiseling operation to a hard tissue using a bone chisel and mallet. In most of the previous studies related to the training simulator, the target organization was soft tissues. However, in oral and orthopedics surgeries, surgeons operate hard tissues such as bone and teeth using a chisel and mallet. Therefore, the development of the surgical training simulator for chiseling the hard tissue by pounding the bone chisel with bone mallet has been promoted in recent year. It is required to represent precisely the chiseling situation for improving the realistic sensation. In this study, we propose the chiseling model with the transition from chiseling to splitting in response to pound the chisel. In the proposed model, the mass, spring and damper model with varying the equilibrium point is applied for representing the elastic and plastic deformation of hard tissue. Moreover, the model parameters are switched for representing the transition from chiseling to splitting. The model parameters are identified by fitting the simulated and the experimented chiseling data. In this study, the measurement of the chiseling motion is also introduced for actual chiseling data acquisition. The efficacy of the proposed model is verified by the comparison with the experimental results from measurement system.

Keywords: Decision Making Systems, Machine Learning, Modeling and Simulating Humans
Abstract: The Wisconsin Card Sorting Task (WCST) is used to measure cognitive flexibility. In WCST, the participants are required to estimate underlying rules (called ”category” in WCST) from given sensory signals. Computational modeling of the underlying cognitive mechanisms of WCST is important for elucidating flexible cognitive processing. In this study, we propose a hierarchical Recurrent Neural Network (RNN) model for explaining the underlying cognitive mechanisms of WCST, based on the free energy principle (FEP). FEP explains perception and goal-directed action as the minimization of prediction errors between predicted and real sensory signals (called free-energy minimization) and is expected to be an integrated theory of the brain. The primary characteristic of our model is that it considers free energy at future time steps, enabling it to correctly answer WCST as a goal-directed behavior based on the FEP. The simulation experiment showed that the proposed model successfully estimated the underlying categories to be estimated in WCST and correctly answer WCST. This indicates that the proposed model may provide mechanistic insights into flexible cognitive processing from the perspective of FEP

Keywords: Machine Learning, Decision Making Systems
Abstract: We envision robots learning new skills as efficiently as possible. A key challenge in this pursuit is that the efficiency of learning systems such as Reinforcement Learning (RL) deteriorates with the task complexity. For instance, in the task of building a tower of cubes, multiple subtasks must be executed in the correct sequence to succeed with this long-horizon task. As the sequence extends, determining the correct order of actions becomes increasingly difficult, particularly for RL methods that rely on trial-and-error. To tackle this, we propose a new method that integrates symbolic models into RL to boost learning efficiency in long-horizon tasks. Symbolic models offer task abstraction and can enhance sample efficiency for RL agents through high-level operators. Our approach focuses on task decomposition aligned with the structure of Automated Planning (AP) operators, enabling RL agents to synthesize individual skills for specific subtasks, thus they will require fewer learning samples. The task decomposition is designed with dual consideration: 1) reducing textit{errors} when linking subsequent skills together and 2) enhancing skill textit{reusability} for downstream tasks with similar structures. In simulated robot manipulation tasks, such as stacking two cubes, experiments demonstrate superior sample efficiency for our proposed approach (a 2x reduction in training cost) compared to most RL baselines. Furthermore, our method is robust to generalise to unseen rearrangement tasks with minimal interaction steps (fewer than 100), achieving an average success rate approximately 50% higher than baselines, which often struggle to make progress.

Keywords: Machine Learning, Human-Robot Cooperation/Collaboration, Rehabilitation Systems
Abstract: Sit-to-stand (STS) motion is a complex behavior and requires a high power, especially when included in activities of daily living. For people with difficulties in performing this motion, assistive devices are particularly suitable. Although assistive devices should timely activate for providing assistance, they often exhibit multiple inherent delays. We introduce a control scheme intended to ensure proper timing for triggering assistive devices by predicting user's STS motion intention. We employ electromyography and muscle synergy and design a deep neural network to predict motion intention. The proposed network is evaluated in terms of the accuracy and anticipation time of the predicted motion according to the forecast time considered during network training. The accuracy decreases with increasing forecast time. In addition, a longer forecast time for training increases the inferred anticipation time. Our results suggest that the control of assistive devices with proper timing and output power is feasible by implementing the proposed control scheme.

Keywords: Human-Robot Cooperation/Collaboration, Decision Making Systems, Virtual Reality and Interfaces
Abstract: Learning from Demonstration (LfD) systems are commonly used to teach robots new tasks by generating a set of skills from user-provided demonstrations. These skills can then be sequenced by planning algorithms to execute complex tasks. However, LfD systems typically require a full demonstration of the entire task, even when parts of it are already known to the robot. This limitation comes from the system’s inability to recognize which sub-tasks are already familiar, leading to a repetitive and burdensome demonstration process for users. In this paper, we introduce a new method for guided demonstrations that reduces this burden, by helping the robot to identify which parts of the task it already knows, considering the overall task goal and the robot’s existing skills. In particular, through a combinatorial search, the method finds the smallest necessary change in the initial task conditions that allows the robot to solve the task with its current knowledge. This state is referred to as the “excuse state.” The human demonstrator is then only required to teach how to reach the excuse state (missing sub-task), rather than demonstrating the entire task. Empirical results and a pilot user study show that our method reduces demonstration time by 61% and decreases the size of demonstrations by 72%.

Keywords: Sensor Networks, Machine Learning, Intelligent Transportation Systems
Abstract: Posture recognition and Heart-Rate measurement towards passenger safety without disturbing the vehicular ergonomics is much appreciated. In this work, we present a novel integrated system utilizing Velostat Pressure Sensitive Conductive (PSC) sensors and an accelerometer designed on a seat-belt. This system captures spatio-temporal data and employs a Convolutional Neural Network (CNN) model for the classification of five distinct sitting postures and non-wearing seat-belt detection. The accelerometer data is processed to estimate Heart-Rate using an algorithm, ensuring accurate and reliable measurements. The non-skin-contact accelerometer unit integrated to the seat-belt and recording heartrate is preferred over other designs owing to the commuters convenience. The inclusion of a diverse subject pool to generate a comprehensive dataset, featuring an unoccupied seat state, ensures robust and realistic performance. The CNN model, interfaced with an edge computing system, with its ability to extract hidden features, achieved high accuracy in posture recognition. Additionally, the camera-less approach preserves user privacy, making it suitable for real-world applications. Our compact seat-belt incorporated sensory system represents a significant advancement in passenger monitoring technology, offering a practical and privacy-conscious solution for improving vehicle safety and ergonomics. The sensory dataset, along with design files, and source-codes are made freely available for further usage to designers and researchers community.

Keywords: Soft Robotics, Mechatronics Systems
Abstract: In this paper, we fabricated a 3-dimensional (3D) organogel by using 3D printing and evaluated its properties. DEAs have been widely used in various fields because of their high energy density and fast response. DEAs consist of a dielectric elastomer and flexible electrodes, and strain of DEAs is influenced by the relative permittivity and Young’s modulus of the dielectric elastomer. Organogel has particularly high relative permittivity and low Young’s modulus compared to other dielectric elastomers. Thus, DEAs made of an organogel can generate significant strain. This characteristic makes DEAs suitable for applications such as organ models and tactile presentation devices. However, it is difficult to fabricate complex-shaped organogels by using conventional fabrication methods. 3D printing is one of the most effective methods for fabricating complex shape. Therefore, we used a 3D printer to realize complex-shaped DEAs made of an organogel.

Keywords: Biologically-Inspired Robotic Systems, Soft Robotics
Abstract: This paper introduces a method to realize stretch reflexes in musculoskeletal robots without length sensors. A pneumatic artificial muscle (PAM) was modeled as a nonlinear spring with a spring constant dependent on deformation and pressure. The dimension of the spring constant had physical bases in prior research, and its coefficients were derived from static tensile tests that measured force and length under a constant pressure. We applied the model to estimate length of four PAMs with different materials and shapes, verifying the general applicability. When incorporated into an arm driven by antagonistic muscles, the model was proved effective in monitoring the velocity change of PAM length and triggering a stretch reflex, enhancing the robot’s adaptability to disturbances. The reflex trajectory with a conventional sensor was well replicated with the model, offering a practical alternative to length sensors.

Keywords: Biomimetics, Biologically-Inspired Robotic Systems
Abstract: For a robot, its skin is an important element that determines its performance and rigidity. Although a wide variety of robots have been actively developed, the performance of their skins is best when they are new. Research has been conducted on self-healing robots, however, their repair capabilities are limited and their performance after repair is no better than their original performance. In contrast, organisms can self-heal and can improve their individual capabilities through growth. Therefore, we aim to incorporate plants into the robot skin to give it different capabilities from conventional robots. In this study, we focused on the rapid growth of plant sprouts and grew pea sprouts and measured their root strength in a tensile test. Furthermore, we connected the two separated sponges by growing pea sprout roots, and confirmed their strength through tensile experimentation. This result indicated the possibility of using plant growth for self-healing capabilities.

Keywords: Soft Robotics
Abstract: Various soft actuators have been developed in the past, achieving flexible motion. However, soft actuators that combine high force and high work have not yet been realized. The ratchet movement between actin and myosin, which causes muscle contraction in living organisms, could potentially lead to such high-efficiency soft actuators. In this paper, we describe a thread-based soft linear actuator inspired by this biological principle. The mechanism of this actuator utilizes the interaction between threads and objects to generate motion. Multiple units could be stacked to produce high force when used as a contracting actuator. To verify the principle of the proposed mechanism, we conducted transport experiments. The object placed on the threads was successfully transported at a speed of 2.2 mm/min.

Keywords: Soft Robotics, Human-Robot/System Interaction, Virtual Reality and Interfaces
Abstract: Postural stability is a significant aspect that needs to be investigated to prevent the risk of injuries and falls in all age groups. Moreover, developing reliable techniques and feedback modalities to enhance the posture control evaluation is crucial. The study investigates the effects of incorporating self-simulated gaming exercises with haptic feedback on posture control in young adults. Participants were randomly assigned to three groups: a control group (n=10) which performed conventional squatting within prescribed time intervals; an intervention group A (n=9), which engaged in virtual reality (VR)-based squatting; and an intervention group B (n=9), which performed squatting in a VR environment while wearing a soft actuator suit. The gaming interface dynamically optimizes the difficulty of squatting based on the user’s knee tremor. We evaluated several center of pressure (COP) parameters: displacement, root mean square (RMS), area, median frequency, and mean velocity in the mediolateral (ML) and anterior-posterior (AP) directions. Statistical analysis using ANOVA revealed significant effects in the ML displacement between the control group and intervention group B (p = 0.027), ML RMS between the control group and intervention group A (p = 0.001), and between the control group and intervention group B (p = 0.001). Additionally, significant differences were observed in the circumference area between the control group and intervention group A (p = 0.035) and between the control group and intervention group B (p = 0.001). The findings indicate that the developed prototype, which provides visual and haptic guidance, shows promise as a tool for improving postural control.

Keywords: Soft Robotics, Mechanism Design, Robotic hands and grasping
Abstract: Soft actuators have many advantages, such as flexibility and safe interaction with the environment. Despite these advantages, they still lack the stiffness to carry the high load. The layer jamming mechanism can be applied to the actuator to increase stiffness, and a vacuum can control the stiffness of the mechanism. However, the typical jamming mechanism has a risk of air leakage, and vacuums are bulky to some systems. In order to solve this problem, we have developed a novel mechanical layer jamming for the cable-driven soft actuator in this paper. Our design uses the actuator body to generate the compression force to the layer at the middle of the actuator’s body to increase the actuator’s stiffness. The result shows the effectiveness of our mechanical layer jamming in increasing the stiffness of the actuator when adding the load to the tip of the actuator. In addition, we have applied our actuator to the soft gripper system, which successfully grasped various objects and changed the stiffness of the actuator.

Keywords: Soft Robotics, Human-Robot Cooperation/Collaboration, Human-Robot/System Interaction
Abstract: In this study, we observe the dynamic behavior of an inflatable robot arm with an internally reinforced drop-stitch structure. We examine the deformation during motion of 1 and 2 degrees of freedom (DOF) for an inflatable body. The inflatable robot arm has a soft inflatable body as links and rigid servo actuators as joints. We implemented a sinusoidal motion for inflatable links for various payload conditions and analyzed them using a Motion Capture system. To estimate the dynamic deformation of the balloon in motion, we have defined a Deformation Index (DI) metric. Angle, current of the actuator (servo), and DI are used as input to polynomial regression to predict the end effector position. With this analysis, we can understand the complexity of modeling the nonlinear behavior of inflatable links for multi-DOF motion. We observed DI helps improve the prediction of the end effector position by including deformation information. However, the results demonstrate the limitations of polynomial regression analysis of an internally reinforced inflatable robot arm link.

Keywords: Mechanism Design, Systems for Search and Rescue Applications
Abstract: Thin and long-reach arms are effective for the decommissioning of the Fukushima Daiichi Nuclear Power Plant, particularly for measuring in-vessel structures and removing fuel debris. Currently, there is a need for an arm that can investigate the bottom of the reactor pressure vessel by attaching a vertically extending arm equipped with investigation equipment to the tip of a horizontally extending long-reach arm. Telescopic arms are expected to provide a high extension-to-contraction ratio, reduce the installation space of the arm, and enable entry into narrow spaces and wide-area observation. In this paper, we propose an extension/contraction mechanism that uses a helically grooved flexible conduit as a feed screw. Using the proposed mechanism, the conduit is fed into an elastically bendable arm, and a prototype that can extend/contract up to 8.6 m is developed. The fundamental performance of the elastic telescopic arm was evaluated through experiments in which the arm was extended/contracted horizontally (0 deg), tilted (45 deg), and upward (90 deg), payload experiments, and an external environment interference experiment.

Keywords: Systems for Service/Assistive Applications, Human-Robot/System Interaction, Rehabilitation Systems
Abstract: In modern society, where the population is aging, fall prevention is an extremely important issue for extending healthy life expectancy, preventing nursing care, and reducing the economic burden on families. However, while numerous studies have been conducted on fall risk estimation, fall motion detection, and fall occurrence detection, studies on fall predic tion remain limited. This is because it requires multiple sensors and wearable devices, which lacks practicality. In this study, we investigated the possibility of predicting stumbling during the initial swing phase for the purpose of preventing falls during walking in elderly people. First, using Simscape Multibody, one of the toolboxes of MATLAB/Simulink, we developed a total of 36 models by combining six leg models and six walking models. Next, we attached one inertial sensor to the toe in the simulation and analyzed and compared the data. As a result, it was confirmed that stumbling can be predicted based on the positive or negative slope of angular acceleration in the first half of the initial swing phase just prior to stumbling.

Keywords: Human-Robot/System Interaction, Medical Systems, Mechatronics Systems
Abstract: Scoliosis is a condition characterized by an abnormal curvature of the spine, which may develop idiopathically during growth or as a result of neurogenic or myogenic diseases. As scoliosis progresses, it can exert pressure on internal organs, disrupt cardiac rhythm, and cause breathing difficulties, particularly in the sitting posture. Early detection and consistent use of an orthosis are essential to manage the progression of scoliosis effectively and to correct the patient’s posture. Conventional orthotic devices face challenges in measuring and controlling the pressure applied to different areas of the torso, as well as in adjusting the bracing force in response to changes in symptoms. This study aims to develop a novel system incorporating airbags that can dynamically adjust the bracing force through air pressure modulation. The system is designed to measure and control the corrective force required for scoliosis treatment, in a sitting position, and to integrate with a chair for practical application. Its basic performance was evaluated through functional experiments. We developed a prototype system capable of automatically applying force at three specific points by adjusting the air pressure in airbags using solenoid valves and air pumps controlled by a microprocessor. To assess its performance, experiments were conducted on a mannequin to apply a prescribed corrective force to three lateral points. The pressure range tested was between 30 kPa and 80 kPa, with increments of 10 kPa. The results demonstrated that the system could reliably apply the prescribed pressure to the mannequin at all tested levels, with an error margin comparable to that of existing adjustable orthotic devices. In conclusion, the experiments confirmed the basic performance of the developed system for scoliosis treatment, indicating its potential utility in clinical applications.

Keywords: Hardware Platform, Mechatronics Systems, Medical Systems
Abstract: In this paper, we introduce an academic-level, highly modular leg prosthesis based upon the Open Source Leg (OSL),2.0, targeting improved manufacturability, cost-effectiveness, and functionality in prosthetic technology. The OSL's structure has been amended by implementing a modular design and more standard components to support flexible configurations for both left and right leg amputations. With respect to the OSL,2.0, our results demonstrate a 40-70% reduction in manufacturing time and costs while maintaining high standards of safety and functionality as confirmed by Finite Element Analysis (FEA), which showed an enhanced safety factor of 1.107 under static loads. The new prosthetic leg achieves a significant cost reduction and facilitates greater accessibility and adaptability, promoting further adoption.

Keywords: Systems for Service/Assistive Applications, Robotic hands and grasping, Mechanism Design
Abstract: This paper presents a robotic system designed to assist hemiplegic individuals in dressing, specifically focusing on the process of putting on bottoms. Assisting a hemiplegic patient with this task involves several complex actions, such as maneuvering through tight spaces and coordinating simultaneous movements with both hands. In this study, we analyze the assistive techniques used by caregivers and propose a branching arm robot equipped with two end-effectors to address these challenges. The proposed robot has a sufficient range of motion and torque, enabling it to assist in dressing while maintaining a compact design. Additionally, it includes a specialized end-effector capable of performing a multi-stage release of the bottoms, accommodating the intricate handling of clothing. Experimental results demonstrate that the robot effectively aids hemiplegic individuals in dressing, from threading the legs through the hems to pulling the bottoms up to the waist.

Keywords: Haptics and tactile sensors, Human Interface, Virtual Reality and Interfaces
Abstract: Traditionally, to learn a sport effectively, people receive feedback from an expert, such as an instructor. However, personal instructors are not only expensive, but the demand for training is also larger than the number of instructors available. As an alternative to traditional instruction, we are developing a sports training system to provide multimodal feedback to users based on their performance and environment information. The feedback is based on both the user's performance and the motion of expert athletes, enabling users to train effectively. Initially, we chose Golf as a case of study to implement this system. In this paper, we introduce an approach to train putting tempo and stroke length using vibrotactile feedback in order to enable users to train their putting ability consistently. Using the proposed vibrotactile cues, users were able to perform putts with tempos close to the ones provided by the vibrotactile cues. Furthermore, they were able to control the length of the backswing stroke following the proposed vibrotactile cues. We believe the proposed framework can be used to train the basic skills of putting effectively.

Keywords: Human Interface
Abstract: A baby bouncer is a type of cradle that allows for rocking and soothing through the baby's own movements or the use of hands by an adult. An electric baby bouncer is designed to automatically rock the baby, enabling parents to soothe their child without using their hands and reducing the effort required. However, existing electric baby bouncers have the drawback of being expensive, heavy, and bulky. In this study, we propose an external attachment device that can be easily added to existing manual baby bouncers to enable automatic rocking. This device operates by transmitting the rotation of a servo motor to the baby bouncer, facilitating the rocking motion. Through experimentation, we validated that the device can adapt to variations in the baby's weight and changes in the servo motor's rotation cycle, thereby allowing for adjustments in the swing range of the baby bouncer.

Keywords: Human-Robot/System Interaction
Abstract: Seventeen percent of the Japanese workforce performs physical labor, and there is a need to reduce the physical burden of this work. Assist devices using artificial muscles, which are lightweight and easy to install, are effective in reducing the burden of physical labor. However, when an assist device is used for lifting a load, for example, the assist torque must be quickly raised to a certain level, which may lead to discomfort for the user, reduced operability. Therefore, this research conducted an experiment to verify whether controlling the assist torque of an assist device using artificial muscles according to the previously measured exerted muscle strength pattern of the lifting motion can reduce the mental load compared to a case where the assist torque is suddenly raised. Participants in the experiment were asked to receive assistance from the assist device, and the results were evaluated based on subjective evaluation and the sum of squares of the joint angular jerk, an index of discomfort. The results showed that there was no significant difference in subjective mental load, but the sum of squares of joint angular jerk decreased significantly.

Keywords: Human Factors and Human-in-the-Loop, Rehabilitation Systems, Welfare systems
Abstract: Isokinetic training leads to rapid skeletal muscle hypertrophy, while slow-speed resistance training is effective for promoting muscle hypertrophy. It is essential to use an effective resistance load that minimizes injury risk in skeletal muscle hypertrophy interventions. Previous studies have indicated that surface electromyogram (EMG) measurements are useful for detecting the load threshold (minimum effective load). However, the attachment position of the electrodes affects the characteristics of the EMG signal readings. It remains unclear whether the system for detecting the load threshold functions correctly when the electrode attachment position shifts. The objective of this study is to explore how the attachment position of EMG electrodes affects the detection of the load threshold. We hypothesized that frequency analysis of EMG signals might be a reliable method for checking the electrode attachment position, given the alternating current nature of EMG signals. As a result, as long as the EMG electrode is placed on the muscle, slight positional shifts are not very critical for detecting the load threshold. In addition, results show that we can verify whether the electrode attachment position is appropriate for detecting the load threshold and make adjustments by examining the magnitude data of the EMG signal frequency bands.

Keywords: Intelligent Transportation Systems, Systems for Service/Assistive Applications, Human-Robot/System Interaction
Abstract: We propose a “surrogate shopping system” allowing users to shop in stores while performing tasks at home. A mobile robot with a robotic arm and camera enables users to view both home and store environments via a head-mounted display, performing household and shopping tasks simultaneously. We compared two interaction methods: a master-slave arm manipulation and a laser pointer-based item selection. The study showed a preference for the pointing method due to its usability in a dual-task situation, though the master-slave method allowed for more detailed shopping.

Keywords: Systems for Service/Assistive Applications, Hardware Platform, Integration Platform
Abstract: This short paper with a video demonstration presents an add-on adaptive speed control module designed for pipeline inspection gauges (PIGs). The module integrates multiple sensors, a rotary valve unit with an actuator, and an embedded controller. The module’s speed control is realized by adjusting the valve angle (morphological adaptation), which modulates bypass flow and, subsequently, moving speed. An Extended Kalman Filter (EKF) is utilized for sensor fusion and module speed estimation. Closed-loop Proportional–Derivative (PD) control automatically adjusts the valve angle or adapts the module morphology based on the difference between estimated and desired speeds, ensuring precise speed regulation. The module’s performance is evaluated and demonstrated in an industrial pipeline environment under various fluid flow rates.

Keywords: Haptics and tactile sensors, Robotic hands and grasping, Virtual Reality and Interfaces
Abstract: We propose a low-cost robot operation interface that combines a mixed-reality device and vision-based tactile sensor for simultaneous grip force fine-tuning and position control. In a paper creasing demonstration, we confirm that the proposed system dynamically reproduces a small force input by the operator to the controller as the grasping force of a robot gripper.

Keywords: Vision Systems, Environment Monitoring and Management, Systems for Field Applications
Abstract: Although the use of a traveling monitoring system to inspect infrastructure structures improves efficiency, there are concerns about image quality degradation and inspection accuracy due to motion blur caused by movement. Therefore, this research aims to develop a method to rotate a transparent acrylic hexagonal prism and control the optical axis in accordance with Snell's law to capture images while compensating for the motion in real time. In the previous study, the same incident angle was visited only every quarter of a rotation because a cube was used, and the overlap region was not included in the continuous image in an optical system with a narrow angle of view. Experiments have confirmed that it is possible to capture continuous images of a moving object in a similar shape at a maximum speed equivalent to 70 km/h, while compensating for motion blur of the object without interruption.

Keywords: Energy and Environment-Aware Automation, Automation Systems, Integration Platform
Abstract: Our research aims to develop an intelligent control system for optimizing the operation of lighting systems for a jogging track in the countryside by using adaptive control methods, artificial intelligence algorithms and optimization of built-in lighting control. Through this, lighting systems will be optimized, energy consumption will be minimized and the lighting system will be adapted to be able to handle a larger amount of local renewable energy production. Due to the level of complexity, in this paper, we have focused to integrate a commercial lighting control system and the solar energy system with energy storage. In particular, their respective components were selected and integrated by means of a distributed control architecture. An API was developed by using the design science research methodology aiming at bridging proprietary systems for sustainable lighting solutions.

Keywords: Systems for Field Applications, Integration Platform, Soft Robotics
Abstract: This video demonstration paper presents an integrated adaptive system for on-pipe landing and adherence of autonomous inspection drones. The system employs a dual-level adaptive method to address the challenges of the stable drone approach, landing, and adherence to pipes for performing subsequent close inspection. The first level, passive adaptation, is realized through morphologically computed double SOft Fin-ray landing gEAR (SOFEAR). It enables the drone to maintain and adhere to the pipe without significant control effort after landing. The second level, active adaptation, achieved via adaptive stability control, facilitates the precision approach and landing of the drone while compensating for external disturbances. The proposed system, incorporating both passive and active adaptation, demonstrates robust drone performance in the real world, including accurate pipe centering and approach, autonomous recovery from off-center landing attempts due to unexpected disturbances, soft landings, and steady on-pipe adherence.

Keywords: Medical Systems, Systems for Field Applications, Machine Learning
Abstract: Blood pressure (BP) is a vital parameter in medical treatment and diagnosis, and as a non-invasive method to measure BP, some deep learning models have been proposed to estimate BP from photoplethysmograms (PPGs). However, the datasets and the method of dividing them into training and testing subsets are not uniform, making it difficult to compare them fairly. In this study, we compared the performance of nine deep learning models for estimating systolic and diastolic BP using PPGs. We used PulseDB, which has "calibration-based subset'' and "calibration-free subset'' as test subsets. The calibration-based subset has the same subject's data in the training subset, and the calibration-free subset does not have the same subject's data in the training subset. The results showed that ST-ResNet performed the best, and it is important to evaluate the models using both calibration-based and calibration-free subsets, and to prevent the overfitting by using the weight decay.

Keywords: Automation Systems, Environment / Ecological Systems, Environment Monitoring and Management
Abstract: There are 50 billion pieces of litter in the U.S. alone. Grass fields contribute to this problem because picnickers tend to leave trash on the field. We propose building a robot that can autonomously navigate, identify, and pick up trash in parks. To autonomously navigate the park, we used a Spanning Tree Coverage (STC) algorithm to generate a coverage path the robot could follow. To navigate this path, we successfully used Real-Time Kinematic (RTK) GPS, which provides a centimeter-level reading every second. For computer vision, we utilized the ResNet50 Convolutional Neural Network (CNN), which detects trash with 94.52% accuracy. For trash pickup, we tested multiple design concepts. We select a new pickup mechanism that specifically targets the trash we encounter on the field. Our solution achieved an overall success rate of 80%, demonstrating that autonomous trash pickup robots on grass fields are a viable solution.

Keywords: Intelligent Transportation Systems, Decision Making Systems
Abstract: This research proposes a method to generate arbitrary activity data for social simulation by utilizing a behavior choice model, learned from synthetic population data and activity data derived from public statistics without relying on hard-to-obtain personal data (actual behavior data). The effectiveness of the proposed method is validated using the activity simulator ActivitySim.

Keywords: Human-Robot/System Interaction, Human-Robot Cooperation/Collaboration, Software Platform
Abstract: Designing service robots capable of tidying up in unfamiliar and dynamic human environments presents a significant challenge. Such robots must not only recognize and manipulate a wide range of objects but also align their actions with tidying up rules, which may vary greatly from one individual to another. To address these challenges, we propose a comprehensive software framework that integrates Large Language Model (LLM) and Vision-Language Models (VLMs) for service robots. Our framework enables robots to learn human-specific tidying up rules through interaction and observation, and to identify and handle previously unseen objects and receptacles. This adaptive framework offers a unified solution for recognizing, learning, and acting upon diverse and dynamic human environments. We evaluate our framework using both a text-based benchmark dataset to assess tidying up rule learning and a simulated environment to demonstrate practical tidying up performance. In the evaluation using the text-based benchmark dataset, our framework selects appropriate receptacles for unseen objects with high accuracy (87.4%), including unseen receptacle categories. The simulation evaluation confirms the effectiveness of our framework in realistic environments and scenarios. This research advances the field of service robotics by presenting an integrated software solution that leverages LLM and VLMs for more personalized and adaptable robot behavior in real-world tasks.

Keywords: Machine Learning, Robotic hands and grasping, Human Interface
Abstract: The integration of language instructions with robotic control, particularly through Vision Language Action (VLA) models, has shown significant potential. However, these systems are often hindered by high computational costs, the need for extensive retraining, and limited scalability, making them less accessible for widespread use.

In this paper, we introduce SVLR (Scalable Visual Language Robotics), an open-source, modular framework that operates without the need for retraining, providing a scalable solution for robotic control. SVLR leverages a combination of lightweight, open-source AI models including the Vision-Language Model (VLM) Mini-InternVL, zero-shot image segmentation model CLIPSeg, Large Language Model Phi-3, and sentence similarity model all-MiniLM to process visual and language inputs. These models work together to identify objects in an unknown environment, use them as parameters for task execution, and generate a sequence of actions in response to natural language instructions. A key strength of SVLR is its scalability. The framework allows for easy integration of new robotic tasks and robots by simply adding text descriptions and task definitions, without the need for retraining. This modularity ensures that SVLR can continuously adapt to the latest advancements in AI technologies and support a wide range of robots and tasks.

SVLR operates effectively on an NVIDIA RTX 2070 (mobile) GPU, demonstrating promising performance in executing pick-and-place tasks. While these initial results are encouraging, further evaluation across a broader set of tasks and comparisons with existing VLA models are needed to assess SVLR’s generalization capabilities and performance in more complex scenarios.

Keywords: Machine Learning, Biologically-Inspired Robotic Systems, Motion and Path Planning
Abstract: Redundant robots with more degrees of freedom than necessary for given tasks have attracted attention due to their flexibility, but they also increase the complexity of control. Especially for highly redundant robots, accurate motion planning and obstacle avoidance remain challenging. This research aims to develop a redundant robot arm that can perform reaching tasks while avoiding randomly appearing obstacles using reinforcement learning. We adopted the Proximal Policy Optimization (PPO) algorithm and conducted simulations in the Mujoco environment. The learning process consisted of a three-stage curriculum: reaching task, fixed obstacle avoidance, and random obstacle avoidance, gradually increasing difficulty to achieve efficient learning. Experimental results showed that the arm could adapt to complex environments and effectively reach target positions while avoiding obstacles. In particular, the system demonstrated high adaptability to randomly placed obstacles, successfully reaching within a maximum distance of approximately 0.07 m from the target position.

Keywords: Motion and Path Planning, Machine Learning, Human-Robot/System Interaction
Abstract: Optimizing a robot's posture can be advantageous for managing interaction forces with the environment. By optimizing the Reflected Mass (RM) along entire trajectories, the robot's posture can be adjusted to minimize impact forces for safety or maximize them for tasks that require high force, such as pushing or striking. However, the integration of RM optimization within motion planning remains under-explored. To address this, we introduce two new approaches for optimizing RM in motion planning: a probabilistic generative model based on diffusion techniques and a sampling-based method using Rapidly-Exploring Random Trees Star (RRT*). Both methods optimize the RM within the motion planning framework, enabling new strategies for enhancing robot interactions in diverse and dynamic environments. Experimental validation in simulation and on a UR5 robot demonstrates the effectiveness of these approaches in controlling RM, offering promising directions for future research and applications.

Keywords: Soft Robotics, Biologically-Inspired Robotic Systems, Environment / Ecological Systems
Abstract: Moving through the ground with a soft robot is a difficult task. Soft locomotion can move without damaging the environment, such as tree roots, but even in just a few centimeters of soil, high friction and resistance forces occur. Differences in soil topography and moisture content also affect the motion. We therefore developed a small, bellows-shaped earthworm robot with an outer diameter of 12 mm. The robot consists of silicone rubber and shape memory alloy wire, and the inside of the bellows is filled with air. When electric current is applied to the shape memory alloy wire, the convex part of the bellows contracts and stretches in the axial direction, generating a force for movement. When no current is applied, it is used as an anchoring segment. We have experimented with 16 different patterns of soil topography and moisture content, and succeeded in realizing soft-robotic subterranean locomotion.

Keywords: Robotic hands and grasping
Abstract: In-hand manipulation with multi-fingered robotic hands is often limited by the restricted range of finger postures that can be achieved when handling objects. Expanding this range is crucial for enhancing the dexterity and versatility of robotic hands in various applications, such as assembly and precision tasks. One of the main limitations comes from the structural design of the fingers, which are typically composed of linked mechanisms comprising fixed-length links and rotational joints. To address this issue, this paper proposes the use of continuum robots for the fingers of multi-fingered hands. We detail the design of the robotic hand and its kinematic model using a constant curvature model, followed by simulations of finger reach and actual grasping experiments. The results of the grasping experiment show that the proposed robotic hand significantly expands the graspable range compared to conventional two-link robotic hands, due to its ability to greatly extend, retract, and bend.

Keywords: Sensor Fusion, Modeling and Simulating Humans
Abstract: This paper describes a method for accurately estimating gait trajectory. We have been developed an estimation method based on the observation from two inertial motion units (IMUs) attached on both insteps and LiDARs equipped on the back. One advantage of the method is no need to observe a person from a fixed point. However, there is room for further improvement of the accuracy of the estimated gait trajectory. In particular, the geometry of the gait was rather corrupted. In this study, we propose the introduction of a passive dynamic gait (PDW) model to maintain the gait geometry, and present the formulation for embedding it into the previous method. We also present a method to complement the discontinuity of the trajectory by PWD. The effectiveness of the method is explained based on the experimental results using real data.　

Keywords: Rehabilitation Systems, Welfare systems, Medical Systems
Abstract: 筋力低下と歩行障害のある脳卒中患者 足首の背屈が不十分な場合の支援が必要 足底屈曲の動き。この研究は、 足首の動きを補助する無動力歩行補助装置 股関節の動きを動力源として使用します。システムとは異なり、 モーターなどのアクティブアクチュエータに依存して、このデバイスは カム機構とケーブルでエネルギーを変換 股関節の屈曲と伸展の動きから背屈へ 歩行中の足底屈曲トルク。ある研究 健常者を対象に実施された結果、以下のことが実証されました。 このデバイスの支援により、 着地時、足底屈曲時の足首関節の角度 後期スタンスフェー

Keywords: Robotic hands and grasping, Mechatronics Systems, Human-Robot Cooperation/Collaboration
Abstract: The authors have developed a five-finger robot hand with artificial fingers consisting of closed linkages and a Single Planetary Gear System. Recently, we developed a new finger mechanism equipping a hyperextension joint at the fingertip that acts passively by torsion springs. This paper aims to verify how the hyperextension joint allows to control the reaction force vector that the fingertip applies to an external object by simulation study.

Keywords: Mechanism Design, Systems for Field Applications, Hardware Platform
Abstract: Moving along porous membrane walls, such as fabric or mesh, presents a significant engineering challenge for wall-climbing robots. Traditional suction mechanisms, such as suction cups, struggle to adhere effectively to porous surfaces. Additionally, if the wall is non-magnetic, conventional magnetic adhesion methods are also ineffective. While drone-based approaches offer an alternative, there is a risk of damaging the membrane with propellers, and like suction-based methods, maintaining position and posture consumes considerable energy. In this study, we address these challenges by proposing a robotic system designed to traverse non-magnetic, flexible membrane walls. The system comprises a vehicle equipped with three omnidirectional drive units, each with two magnetic wheels, mounted on both sides of the membrane. As a preliminary step, we conduct a fundamental analysis of the vehicle’s configuration, develop a prototype, and validate its basic effectiveness through practical experiments with the physical model.

Keywords: Mechatronics Systems, Control Theory and Technology, Biologically-Inspired Robotic Systems
Abstract: This study discussed a generation of wheel gait for planar 4-DOF compass-like biped robot. Wheel gait is a novel locomotion pattern in which the stance leg swings in the opposite direction to that of a normal gait; that is, both the stance and swing legs rotate in the same direction. We derived dynamic equations of the robot and built an output tracking control of the hip angle. Next, we searched for fixed points using numerical approaches, and revealed that it is possible to generate a steady wheel gait on a slope with appropriate initial states. However, since the limit cycle around the fixed point was identified to be unstable, it is expected that achieving a stable wheel gait with the control method in the real world is nearly impossible. On the other hand, small regions near the fixed point were found where stable period-2 gaits emerge, pointing to the possibility of stable wheel gait generation.

Keywords: Welfare systems, Systems for Service/Assistive Applications
Abstract: This study proposes a novel robot designed for inspecting body surfaces while attached over clothing. This inspection robot consists of a driving unit with a motor and a camera unit with a camera for observing the body surface. The camera unit is placed inside the clothing, while the driving unit is positioned outside, and both units are attached through magnetic gears over the clothing. The motor's driving torque from the drive unit is transmitted to the wheels of the camera unit via the magnetic gears, allowing both units to move in parallel. This enables the robot to move across the clothing and inspect the body surface underneath. The robot is envisioned for applications such as checking the body surface condition of individuals who have difficulty changing clothes. This study also derives the conditions necessary for the robot to move effectively over clothing. The effectiveness of the robot is demonstrated through experiments.

Keywords: Mechatronics Systems, Medical Systems, Integration Platform
Abstract: Fibrosis is a key factor in diseases such as interstitial pneumonia and cirrhosis. Conventional approaches in cell therapy, which involves transplanting living cells into patients, are able to address these conditions. However, cell cultivation involves numerous intricate procedures. Given the demands for a substantial quantity of cells in the therapy, manual cultivation has its limitations. Furthermore, removing cells from a CO2 incubator during culturing can cause damage to the cells, leading to reduced cell viability. To improve the efficacy of cell therapy, it is crucial to cultivate cells with minimal damage. The purpose of this study is to develop a robotic cell culture system that enables the automatic cultivation of fibroblast cells in large quantities for demands of cell therapy, in an environment similar to that of a CO2 incubator and allows to monitor the culture environment. The system enables observation of cells under a microscope without altering the culture conditions and automates the medium exchange process. To confirm the feasibility of the system, we conducted the basic experiment set comparing the developed system to manual cultivation of fibroblast cells. The results showed the developed system improved the fibroblast viability and proliferation rate compared with manual cultivation. In conclusion, we confirmed the feasibility of the developed robotic cell culture system in its capability for cultivating fibroblasts with an efficiency equivalent to or greater than conventional manual cultivation.

Keywords: Medical Systems, Welfare systems
Abstract: Anomalies in motor function and social communication skills constitute early signs of autism spectrum disorder (ASD). In recent years, researchers have paid attention to the possibility of the quantitatively evaluating infants' social communication skills based on their eye movement. In this paper, we attempt to evaluate ASD risk using infants' eye movement data by examining the relationship between eye movement patterns at 4, 10, and 18 months of age and ASD risk as assessed by caregivers using a checklist at 18 months of age. A total of 18 eye movement indices were calculated from eye movement data measured by an optical eye tracker. The results revealed an association between ASD risk and eye movement indices. The indices included those related to a preference for social information at 4 months of age, those related to visual exploration of images at 10 months of age, and those indices related to both preference and visual exploration at 18 months of age. These results indicate that eye movement analysis may potentially serve as a means for early screening of ASD risk.

Keywords: Medical Systems, Systems for Service/Assistive Applications
Abstract: To improve the efficiency of surgical skill transfer and proficiency in laparoscopic surgery, an objective method of quantifying surgical skills was developed. Surgical trainings for laparoscopic nephrectomy were conducted using cadavers, and the movements of surgical instruments were measured using Mocap system in 46 trainings. Because the surgical techniques required are different depending on the surgical process, the entire nephrectomy was divided into three surgical processes (Part 1: colon mobilization, Part 2: renal vascularization, and Part 3: dissection of the remaining tissue), and skill analysis was performed for each process. Surgeons were categorized into three groups according to the number of surgical experiences (novice: 0-9 surgeries, intermediate: 10-49 surgeries, and experts: 50 or more). Three-group tests were conducted on 111 features extracted from the surgical instrument movement data. The three-group tests showed that there were significant differences among the three groups in the surgical processes: part 1, part 2, and part 3. Principal component analysis (PCA) was conducted using the features that showed significant differences in the efficiency of the operation.

Keywords: Medical Systems, Welfare systems, Machine Learning
Abstract: Early detection of infants with autism spectrum disorder (ASD) can lead to effective developmental support. In clinical practice, early screening of 18-month-old infants is implemented using a parent-completed questionnaire. However, research has suggested that signs of ASD may also appear in movement characteristics during the first few months of age. In this paper, we propose a method to evaluate infant movement from videos based on non-negative tensor factorization (NTF) and apply it to ASD risk assessment in the neonatal period. The proposed method applies NTF to pose data estimated from videos, and decomposes the infant's movements into multiple components while considering the linkage between body parts. In the experiment, we evaluated the effectiveness of the proposed method using 36 low-risk infants and 13 high-risk infants for ASD, with the aim of applying the method to ASD risk assessment. The results showed that the proposed method captured the tendency for infants to perform different movements depending on their risk level. Machine learning analysis revealed that the proposed method identified ASD risk with an accuracy exceeding 70%, which was comparable or superior to the existing video evaluation method based on heuristically designed indicators.

Keywords: Medical Systems, Soft Robotics, Mechanism Design
Abstract: This study presents the first prototype of an insertion mechanism designed to reduce pain during cystoscope insertion. Traditional cystoscopy often causes discomfort as the cystoscope rubs against the urethral wall and its tip impacts the curved urethra. To address this issue, this study focuses on developing a “painless” automatic insertion mechanism for cystoscopy in the urethra, prostate, and bladder using a soft growing actuator. The paper introduces a compact soft growing actuator made from a low-friction material capable of inverted growth. The prototype actuator extended to a maximum length of 140 mm in just 0.10 seconds, with no significant variation in the elongation-time relationship when air pressure was applied at 70–90 kPa. The prototype successfully guided a cystoscope like wire from the urethral entrance to the bladder within an S-tube that replicates the scale and curvature of the human urethra. These findings suggest that the proposed cystoscope insertion mechanism, utilizing a soft growing actuator, has the potential to enable automated cystoscopy for rapid and painless examination of the human urethra.

Keywords: Medical Systems, Systems for Service/Assistive Applications, Machine Learning
Abstract: Endoscopic sinus surgery (ESS) is a standard procedure performed worldwide for nasal and paranasal sinus diseases, but it requires a high level of skill, making efficient training essential. A system for evaluating the skill level of surgeons using machine learning based on measurement data from endoscopic sinus surgeries performed with a 3D sinus model has been developed. In this study, an analysis using SHAP values was conducted to investigate behaviors that significantly contribute to skill evaluation. Furthermore, classification accuracy of the machine learning-based skill classification system was attempted to be improved by reducing features with low contribution. The skill was analyzed using the features in the model that achieved the highest accuracy after feature reduction.

Keywords: Medical Systems, Automation Systems, Systems for Field Applications
Abstract: While administering cancer vaccines, an extra-fine needle insertions into the affected area. Reportedly, needle insertion speeds affect the insertion reaction force, causing needle deflections. Additionally, longer insertion times increase the effect of respiratory fluctuations on the needle insertion speeds, leading to a reduction in needle deflections. This study focused on deriving the optimal speed for the liver, kidney, fat, and muscle of pigs to enable efficient and accurate insertion. Insertion tests were conducted to analyze the dependence of needle deflections on the insertion speed ranging from 2 to 20 mm/s and insertion angles of 0°, 15°, 30°, and 45°. Subsequently, a diagram of the relationship among insertion speeds, insertion angles, and needle deflections was created for each organ; this illustration, suggests that it is possible to determine the optimal insertion speed. In future, it is necessary to derive the optimal insertion speed for organs not used in this study. It is desirable to use insertion planning in combination with organ segmentation techniques using computed tomography (CT) images to optimal insertion speed.

Keywords: Medical Systems, Mechatronics Systems
Abstract: Fine-needle aspiration cytology (FNAC) is an inexpensive, quick, and minimally invasive diagnostic method for cancer. Physicians need to be highly skilled to perform FNAC as they are required to collect the appropriate quality and quantity of specimens. Many research have reported the robotic needle insertion method for improving the accuracy of needle placement using fine needles. These methods can expand the applicability of FNAC to organs deeper in the body, enabling less-invasive FNAC. However, there is little research focusing on the suction conditions and quality and quantity of specimens with extra-fine needles. In this study, we analyzed the suction conditions for reducing blood contamination and securing the necessary collection amount. The magnitude and duration of suction were chosen as the suction condition parameters. The effect of these parameters on collection amount and blood contamination was experimentally analyzed using pig organs. Blood contamination was evaluated by injecting ink into the organs to simulate blood and measuring the mass ratio of the liquid components of the specimen. The results showed that a long suction duration was effective in increasing the collection amount, and the collection amount reached saturation. Conversely, a small magnitude of suction was effective in reducing blood contamination in the specimens.

Keywords: Mechanism Design, Factory Automation, Mechatronics Systems
Abstract: In this study, we measured the natural frequency and damping ratio of onyx-material robot arm specimens and obtained the variation law of the Young's modulus and attenuation coefficient with filling density. Based on this, we obtained the equations for estimating the natural frequency and damping ratio of robot arm specimens for any filling density and size. This equation allows estimating its vibration characteristics even without printing the robot arm. Finally, we combined the equations for calculating the natural frequency and damping ratio with those for the settling time and investigated how to determine the optimal filling density under different loads. The results show that the anti-vibration effect of the specimen is the worst for a filling density of 25-30%. Therefore, this filling density should be avoided. We also found that the settling time for a 10% filling density is shorter than that of many specimens with a high filling density. To achieve light weights, a 10% filling density may be an option. The proposed equations and curves can provide guidelines to design a robot arm with anti-vibration effect.

Keywords: Mechanism Design, Mechatronics Systems, Hardware Platform
Abstract: In recent years, the demand for industrial robots has increased and is expected to continue to grow. On the other hand, efforts to reduce the global environmental burden are spreading. This paper aims to develop a lightweight reduction gear for energy-saving robots. To reduce the weight of robots, it is necessary to replace conventional metal materials with new lightweight materials. However, in general, weight reduction tends to reduce the rigidity of the robot, resulting in a decrease in the high-speed and high-precision performance required of robots. The authors have studied the possibility of replacing metal parts with machined CFRP, POM, and 3D printer resin parts , and have examined the adaptability of these parts. In the previous paper, weight, no-load running torque, torque-torsional characteristics, and static torque transfer efficiency were compared for 14 different combinations of metal and resin reduction gears. In this paper, the number of reduction gear combinations is increased to 20, and dynamic torque transfer efficiency is added to the comparison. As a result, the effect of the type of resin applied and its combination on the performance of the reduction gears is clarified, and it is shown that weight reduction can be achieved while minimizing performance degradation by selecting the combination that best suits the application.

Keywords: Mechanism Design, Mechatronics Systems, Hardware Platform
Abstract: The aim of this study is to establish a lightweight construction method for robot elements using metal and plastic materials in order to achieve energy saving in industrial robots. However, in general, weight reduction reduces stiffness, which in turn reduces the performance required for industrial robots. This paper presents an experimental setup and models to investigate the effects of link stiffness and joint stiffness on overall stiffness, and demonstrates the validity of the models through experiments. Then, experiments with different link materials were conducted and the results were compared. The influence of the fastening method and the fastening torque of the link on the stiffness of the entire system was clarified.

Keywords: Mechanism Design, Hardware Platform, Factory Automation
Abstract: Replacing the metal structural parts with plastic structural parts will directly reduce energy consumption, thanks to their lightweight. However, plastic materials are prone to creep deformation at room temperature, potentially leading to decreased end-effector positioning accuracy during prolonged use. Previous measurements of creep deformation in lightweight plastics revealed significant deformation in 3D printed parts, likely due to low infill rates and suboptimal reinforcement fiber placement. This paper presents the results of creep deformation measurements on test specimens with improved structural designs, including modified infill rates and fiber placements. The experimental results show that continuous carbon fiber placement along the outer wall direction yielded the least creep deformation. Additionally, variations in infill rates had minimal impact on creep deformation.

Keywords: Haptics and tactile sensors, Soft Robotics, Sensor Fusion
Abstract: Conductive polymer composites with piezoresistive properties have found extensive use in force sensing applications, such as strain gauges. Polylactic acid filled with carbon black nano-fillers is one such easily available and inexpensive material. The piezoresistive properties of conductive polymers have been shown to share an equivalence with the linear viscoelastic relationship between stress and strain. In this study, we examine six progressively complex resistive-viscoelastic models based on this relationship. After empirically determining the model parameters, we compared their stress prediction outputs using resistance data as input, which had been recorded while subjecting a 3D-printed conductive solid to a compressive force. We observed the model performances for two cases: when model parameters are determined from the complete test cycle and when they are determined phase-wise. Phase-wise coefficients showed a much better prediction accuracy and it was also observed that simpler models gave a better performance than more complex models.

Keywords: Mechanism Design, Hardware Platform, Factory Automation
Abstract: Several challenges arise when considering the application of 3D-printed parts to robotics. First, there is less discussion on bending strength. Second, variations in bending strength occur depending on the type of 3D printer used. Third, the fabrication of large-scale parts remains difficult. In this study, to address the first challenge, we conducted three-point bending tests on specimens with dimensions deviating from standardized specifications, revealing discrepancies in material properties and emphasizing the importance of filament crystallization. For the second challenge, we fabricated identical test specimens under the same conditions using two different 3D printers. We performed three-point bending tests, thereby elucidating the variability in bending strength attributable to the printer type. To address the third challenge, we proposed a joining method involving the mating and adhesion of square-pyramidal convex/concave parts and demonstrated its effectiveness through three-point bending tests.

Keywords: Mechanism Design, Mechatronics Systems, Multi-Robot Systems
Abstract: 3D printed parts have anisotropic characteristics depending on the infill structure, and strength changes with infill rate. Recently, some researches have being conducted to elucidate their characteristics, and it is expected that 3D printed parts will be used as robot parts. However, since the size of the 3D printed parts depends on the size of the printers. It has been required to combine to make 3D printed parts. There are few strength evaluations that consider the combination of 3D printed parts. In this study, three-point bending experiments were conducted on 3D printed parts when two or more parts were adhered as one assembled 3D printed component, and the changes in strength and stiffness were confirmed to compare with non-bonded parts. In addition, experiments were also conducted for different bonded shapes and infill structures. As a result, the bonded parts exhibited 0.73 times the strength of the non-bonded parts, with no significant change in stiffness. And it was also shown that the strength changes depending on the bonding shape. Furthermore, the strength of the bond also showed increases depending on the infill structure and the infill rate.

Keywords: Decision Making Systems, Human-Robot/System Interaction
Abstract: Open vocabulary task execution is crucial in autonomous robotics, particularly for indoor service robots operating in dynamic, human-centric environments. Conventional dictionary-based approaches either fail to capture the diversity in interactions between objects and humans or often face scalability issues in memory and computation over time. Thus, a framework capable of executing high-level tasks and robust open-set capabilities is desirable. We consider the task of searching for dynamic objects in an indoor environment called Object Search. While the state-of-the-art approaches focus on the most effective ways to search for a closed set of objects, we propose a framework capable of generalizing to unknown, unseen, and ultimately an open set of objects. Our framework consists of a method to leverage priors of a fixed set of objects to generate task-driven priors for an open set of objects. We utilize Large Language Models (LLMs) and fuzzy logic to facilitate this prior generation. Additionally, the proposed framework also captures the physical layout of the environment to inform task-driven prior generation. Finally, we validate our framework through extensive real-world experiments and provide comparisons with competitive methods, demonstrating its effectiveness in generalizing to an open-set of objects. The results demonstrate our framework’s superiority in reducing search time, distance, and number of visited landmarks, outperforming related methods.

Keywords: Human Factors and Human-in-the-Loop, Systems for Service/Assistive Applications
Abstract: To improve the quality of life of elderly people, it is important to maintain or improve their physical performance. Continuous exercise is effective in maintaining or improving physical performance of elderly people. However, continuing exercise requires a high level of motivation, which can be difficult to maintain. We aim to develop a new system that provides appropriate interventions to motivate elderly people to exercise. In this paper, we propose a method for building a personalized exercise cheering system using a large-scale language model. Additionally, we conduct a fundamental investigation by comparing personalized cheering, which reflects the user’s personality, with non-personalized cheering.

Keywords: Sensor Networks, Machine Learning, Decision Making Systems
Abstract: In the context of food spoilage prediction, the accuracy of predictive models is critical for ensuring food safety and minimizing waste. Traditional loss functions often fail to adequately prioritize errors based on the varying significance of prediction intervals. This research introduces a customized loss function - PeCAR (Penalized Conditional Absolute Regularization) loss, tailored to enhance the predictive performance of ANNs for food spoilage time estimation. The proposed loss function incorporates the actual spoilage time of food items, thereby penalizing errors in short-term predictions more heavily than the same amount of error in long-term predictions. The proposed approach ensures that an absolute prediction error is weighted according to the relative importance of the time frame, reflecting the need for precise short-term predictions in items with shorter spoilage times. The result indicates a substantial improvement of 35.78% in Mean Absolute Error (MAE) and 25.32% in Mean Squared Error (MSE), enhancing the reliability of the model's predictions for food spoilage. A unique chamber setup is designed for acquiring the dataset for different sets of food items. The 4-layer ANN model with the dataset of all the food items is made publicly available for easy adoption and further usage by the researchers and developers community.

Keywords: Machine Learning, Control Theory and Technology, Human Factors and Human-in-the-Loop
Abstract: In robot control from demonstrations, a sufficient dataset cannot be collected for many of the tasks that require experts with qualifications to special skills. Unfortunately, insufficient expert dataset would manifest various types of noise hidden in it. Since adding data is difficult as well, offline imitation learning needs to be robust to such a noise. In the conventional work, a behavioral cloning method based on Tsallis statistics has been developed. However, it weights each data with absolute rating with a fixed threshold, which would fail to imitate coarse/diverse motions. Therefore, this paper improves the conventional method by adding the function of relative rating for each data, which should enable robots to imitate non-noisy data even from coarse/diverse motions. This function can be obtained from a different derivation way of the optimization problem with Tsallis statistics. By integrating it with the conventional derivation way, the proposed method can adjust between the absolute and relative ratings. Finally, for more convenience, we design optimization tricks for the hyperparameters to maximize the variance of weights with avoiding extremely large weights. In numerical simulations and real-robot experiments, we demonstrate the robustness of the proposed method.

Keywords: Decision Making Systems, Autonomous Vehicle Navigation, Machine Learning
Abstract: Are we on the verge of developing robots that can reprogram themselves to overcome unexpected situations? As robotic systems and artificial intelligence continue to evolve, the concept of self-programming robots capable of adaptive reasoning seems to be becoming a tangible reality. This paper demonstrates how Large Language Models (LLMs), specifically the OpenAI o1-preview model, can empower mobile robots to autonomously analyze failures and modify their operational code in real time. Traditional robot programming, bound by the need to anticipate all possible scenarios, often leads to rigid behaviors when faced with unforeseen obstacles. We present novel results where a robot, tasked with navigating a predefined path, encounters an unanticipated obstacle and autonomously generates modified code to address the challenge, including obstacle avoidance and dynamic path planning strategies. By utilizing a structured prompt and advanced reasoning capabilities, the robot moves beyond pre-programmed limitations, embodying a new level of dynamic, self-adaptive autonomy. This work highlights the transformative potential of LLMs in robotics, offering a glimpse into a future where robots not only perform tasks but also learn, adapt, and evolve autonomously in complex, real-world environments.

Keywords: Machine Learning, Vision Systems
Abstract: Gait is unique to individuals and can be acquired from a distance, making it difficult to disguise. Gait videos also contain many elements unrelated to gait, which make gait recognition challenging. Departing from common approaches that use preprocessing such as silhouette extraction, the RGB-based method extracts gait features directly from RGB gait videos. RGB-based methods leverage the difference between two inputs with different attributes to separate gait-related/unrelated features, but their separation performance depends on the diversity of the dataset. To increase the amount and diversity of training data, we focus on the latent space of gait-independent features (style features), which are usually not needed for gait recognition. In this paper, we propose S2Gait (Style feature Sampling Gait), which augments the training data online with images generated from gait-dependent features of the input images and sampled style features. Experiments demonstrate that the proposed method surpasses existing RGB-based methods on almost all metrics for both generated image quality and identification accuracy. We also explore the relationship between the amount of data augmentation and performance taking advantage of our method's flexibility to generate a wide variety of gait images.

Keywords: Multi-Modal Perception, Machine Learning
Abstract: This paper investigates the adaptability of Vision-Language Models (VLM) use in environments other than in the West, exemplified by CLIP. While models like CLIP exhibit commendable performance on established image datasets, their effectiveness in recognizing objects within specific cultural contexts remains an open question. Our experiments, conducted in a simulated environment, reveal noteworthy performance disparities between Western and Japanese datasets. Additionally, we explore the integration of a segmentation model to obtain segmentation masks with language-aligned features. By addressing these crucial gaps, our study provides insights into the nuanced challenges of cross-cultural recognition within the vision-language paradigm. These findings contribute to informed and unbiased model development for practical applications across diverse cultural domains. end{abstract}

Keywords: Human Interface, Haptics and tactile sensors
Abstract: Ring-shaped devices are small, socially acceptable wearable devices that are gaining attention as health tracking and input devices. Although various ring-shaped input devices have been developed, they are limited in the direction of rotation when worn, requiring the user to be aware of the wear direction while wearing or operating the device. We propose the SkinRing, a ring-shaped device that enables gesture input on the side of the finger, regardless of the wear direction. The SkinRing consists of eight photo-reflective sensors attached to a ring, which is put on the index finger and acquires skin deformation information for gesture identification when the thumb touches the side of the index finger. The initial calibration of the hand's opening and closing allows for input independent of the direction of rotation of the ring when worn. A user study, where the identification results of nine gestures were given, shows an average accuracy of 87.8%.

Keywords: Mechatronics Systems, Factory Automation
Abstract: Manufacturing defects such as bulging, or overlap can occur when winding copper wires by winding machines. In this paper, the method of dynamic behavior analysis of copper wire is proposed. First, the deformation of copper wire on physical constraints is formulated considering elastoplastic expansion, contraction, bending, and torsional deformation. Next, supplying of the copper wire is modeled to reduce computational load and unnecessary vibration of the copper wire in the simulation. Finally, the dynamic behavior of the copper wire is simulated when it is wound by the winding machine. The simulation results showed that the bulging problem is related to core rotation speed and the tension of copper wire supplied.

Keywords: Mechanism Design, Mechatronics Systems, Hardware Platform
Abstract: The utilization of synthetic fiber ropes in robot driving systems facilitates designs that are more lightweight and compact compared to their stainless steel counterparts. However, rope elongation during operation results in tension loss, necessitating a rope tensioning system to wind the rope and maintain adequate tension. We have previously proposed a tensioning mechanism wherein two shafts are coaxially connected via a one-way clutch, allowing rotation only in the direction of rope tension for fine-tuning the winding length. Nevertheless, this method subjects the one-way clutch to the full torque generated by rope tension. Consequently, if the torque exceeds the allowable limit during locomotion, the one-way clutch may sustain damage due to impact torque. To address this limitation, this study proposes a novel mechanism that integrates a one-way clutch and a hose clamp to enhance the load-bearing capacity of the tensioning mechanism. Furthermore, experimental result suggests that the allowable torque of the tensioning mechanism can be significantly increased by introducing a wire mesh between the hose clamp and the shaft, thereby augmenting the friction coefficient of the fastening component.

Keywords: Network Systems, Mechanism Design, Integration Platform
Abstract: We propose a two-port representation of the inerter and conduct experiments for verification. Considering the mechanical components and the interactions, the model can successfully forecast the inerter responses during experiments, increasing the fitting accuracy from 49% to 75%. The results show that the two-port network can effectively forecast systems' responses employing the inerter.

Keywords: Welfare systems, Human Interface, Rehabilitation Systems
Abstract: Mechanical vibration stimulation to muscles influences human motion through the phenomena of kinesthetic illusion (KI) and tonic vibration reflex (TVR). On the other hand, some people suffer from disability of their motor functions. Trunk flexion and extension are essential movements involved in many daily activities and often play a critical role in coordinating with limb motions. The effects of vibration stimulation on trunk muscles involved in flexion and extension might be applicable for the rehabilitation of the people suffer from disability of their motor functions. This research applies mechanical vibration stimulation to the trunk muscles for the aim of rehabilitation, measuring the resultant KI and TVR in the motion of trunk flexion and extension, comparing the outcomes across different vibration frequencies.

Keywords: Human-Robot Cooperation/Collaboration, Mechanism Design
Abstract: Enhancing back-drivability in robot joints is crucial for safe and effective physical human-robot interaction (pHRI). This paper presents a novel approach to improve backdrivability by reducing reflected inertia through a multi-motor system (MMS). Unlike traditional high-ratio gearboxes, which amplify motor inertia and reduce efficiency, the MMS distributes torque across multiple motors, allowing for lower gear ratios and less reflected inertia. We developed an optimization model considering motor selection, gear ratios, and gear stages to minimize reflected inertia while meeting load and geometric constraints. A case study on an industrial robot’s shoulder joint demonstrates an 88.84% reduction in reflected inertia using the proposed MMS compared to a conventional single-motor system. The findings suggest that multi-motor systems can significantly reduce reflected inertia, improving back-drivability in robot joints for safer and more efficient pHRI.

Keywords: Soft Robotics, Haptics and tactile sensors
Abstract: In order to enhance the capability of soft robot manipulator, we propose Roddy, a new type of soft robot system for manipulating a “rigid” tool softly. In Roddy, an elastic rod with a 6-axis force sensor is inserted in series between a rigid tool and a rigid robot manipulator. By utilizing the real-time shape estimation technology, Roddy can recognize the rigid tool position/orientation relative to a rigid environment with high accuracy and precision. Due to high compliance of Roddy, we can achieve accurate calibration of the manipulator tool relative to the environmental flat plane. Moreover, we can also achieve to map the location of obstacles on a plane by exploratory collisions. Based on this obstacle mapping algorithm, we demonstrate that Roddy can safely achieve quick scooping tasks where the rigid tool of the robot manipulator is continuously contact with the environmental rigid plane.

Keywords: Automation Systems, Mechanism Design, Soft Robotics
Abstract: This paper presents a robotic hand that utilizes both a locking mechanism and the elasticity of silicone belts to handle heavy agricultural products, such as radishes, addressing the labor shortage in fruit selection facilities caused by the declining birthrate and aging population. When grasping heavy agricultural products, significant force and stability are required to securely lift and transport the product. The proposed robotic hand combines a locking mechanism with the inherent elastic force of silicone belts, eliminating the need for active actuators such as motors or air compressors. The locking mechanism is embedded within the joints of the hand, allowing it to open and close through the combined effect of the silicone belts' elasticity and interaction with the environment. First, the effectiveness of this combined approach was validated through experiments involving the handling of heavy vegetables, including radishes and other similarly weighted objects. Next, the characteristics of the selected silicone material were measured, analyzed, and incorporated into the design. Finally, grasping experiments were conducted using the proposed hand, which leverages both the locking mechanism and the elasticity of silicone belts. The results indicate that the proposed robotic hand, which integrates both a locking mechanism and silicone belts, can reliably and stably grasp heavy agricultural products.

Keywords: Human Interface, Systems for Service/Assistive Applications
Abstract: In holding behaviors, the posture of a neonate is crucial for ensuring safety. In this study, assuming a system for users to learn proper holding posture, we propose an approach for measuring the posture of a neonate during holding behaviors. Our method estimates the posture angles of a neonate based on holding behaviors captured by a camera. By using regression models to estimate a neonate’s posture angles based on the posture of a person holding a neonate, estimation can be made of the posture angles of the neonate without directly attaching sensors to an actual neonate. The results show for within-participant validation, both the inclination angle– the angle at which the doll is rotated around its chest from a horizontal position towards raising its head– and the adduction angle– the angle at which the doll is rotated from a horizontal position around its body axis towards the person’s body side– could be estimated with high accuracy, but for between-participant validation, only inclination angles could be estimated.

Keywords: Welfare systems, Systems for Service/Assistive Applications, Human Interface
Abstract: Deaf and hard-of-hearing (DHH) people have difficulty obtaining information from hearing. Therefore, infor mation from other senses, such as sight, plays an important role. However, in everyday life, DHH people face life-threatening challenges, such as the inability to notice approaching haz ardous sounds from blind spots or from behind, and the time it takes them to avoid danger. In this paper, we focus on one of the hazardous sounds, the driving noise of a car, and propose a system to detect the approach and passage of a car to the device. The effectiveness of this system is demonstrated by detecting in real time the approach or passage of a car toward the wearer based on the time variation of the frequency received by the microphone installed on the smart glasses, and displaying the results on the lens. We discuss the proposed method and our validation of determining the approach and passage of cars.

Keywords: Haptics and tactile sensors, Modeling and Simulating Humans, Virtual Reality and Interfaces
Abstract: To display spatial haptic distributions on the skin in virtual reality (VR) applications, it is essential to consider the neural signals generated by mechanoreceptors in response to haptic stimuli when contacting virtual objects. This paper focuses on reproducing the neural activities of Merkel cells, which are known as slowly adapting type-1 (SA1) mechanoreceptors.The firing activity of these receptors is known to respond to strain energy density (SED). We developed a real-time SED calculation method using a specially designed finite element (FE) model of a fingertip, which features a simplified layer structure of skin and nail. We validated the model by comparing its calculation results with those of a detailed model and a homogeneous model, using reliable FEA software.We demonstrated that modeling the layered structure of the skin is effective in improving the reproducibility of the SED distribution at the locations of Merkel cells.

Keywords: Machine Learning, Human Interface
Abstract: We estimated the kinematics of the upper limb during baseball pitching using a sensor-embedded ball. Twenty collegiate pitchers performed 24 trials each, throwing fastballs, curveballs, and sliders. Kinematic data were collected using motion capture system and a sensor-embedded baseball. Machine learning models, including Linear Regression, Lasso, Random Forest, Gradient Boosting, and Support Vector Regression (SVR), were used to estimate joint angles. The evaluation metric was the R² score from 10-fold cross-validation. Random Forest (R2 score (the average of all joint movement) = 0.85) and Gradient Boosting showed high accuracy (R2 score (the average of all joint movement) = 0.85), particularly for shoulder and elbow joints. The ensemble model further improved accuracy. The model demonstrated high accuracy in estimating joint angles for the shoulder joint (external/internal rotation) (R2 score (the average of all time) = 0.97) and elbow joint (supination/pronation) (R2 score (the average of all time) = 0.96). In the future, the application of this model is expected to facilitate the acquisition of kinematic data of the pitching arm in competitive environments.

Keywords: Welfare systems, Systems for Service/Assistive Applications
Abstract: An effective bottle-feeding method for infants with cleft lip and/or palate (CLP) is the Kumagai method. However, the number of specialists in the method is limited and providing direct instruction to many people is challenging. To solve this problem, developing a guidance system that enables learners to acquire Kumagai method skills without a direct instructor is desirable. Constructing such a guidance system requires defining a quantitative assessment index for skills in the Kumagai method. Therefore, in this study, we propose a new quantitative assessment index for infant-holding posture, which is the most basic and crucial skill in this method. Our experiments confirmed that the proposed index could stably assess the reproducibility of the infant-holding posture in the Kumagai method while reducing the effect of postural fluctuations during the feeding motion.

Keywords: Welfare systems, Human Interface, Rehabilitation Systems
Abstract: The tonic vibration reflex (TVR) phenomenon refers to the sustained involuntary contractions induced in skeletal muscles when subjected to mechanical vibration stimulation. Thus, utilizing vibration stimulation to induce involuntary limb movements promises to become a novel technique for human motion adjustment based on external devices. In recent years, an increasing number of studies have focused on how to apply TVR in welfare domains, such as tremor suppression and rehabilitation training. However, the characteristics of the TVR induced by vibration stimulation may vary across different muscle states, and changes in TVR effects under conditions of muscle fatigue have not been systematically investigated. In this study, by applying mechanical vibration stimulation to the tendon of origin in biceps brachii, the primary muscle in elbow flexion movements, the changes in the effects of TVR induced by vibration stimulation under muscle fatigue conditions are compared. Through experimentation, this study confirmed the efficacy of vibration stimulation-induced TVR in producing limb movements and investigated the variations in reflex intensity under different muscle fatigue states.

Keywords: Rehabilitation Systems, Human-Robot/System Interaction, Medical Systems
Abstract: The effectiveness of lower limb mobility assistance heavily relies on accurate gait pattern detection. Optimal support is achieved when sensors reliably detect gait information. This study aims to evaluate the accuracy of gait event detection using force plates compared to in-shoe IMUs under real-time conditions. Despite favorable sensor placement for real-time data collection, several challenges arose in assessing the system’s accuracy. Four healthy adults participated, performing self-paced walking on force plates while wearing in-shoe IMUs. Sagittal angular velocity (ωsagittal) is the rate of rotational movement around the sagittal plane. As these measures demonstrated the most importance to accurate event identification, we concentrated on them in our investigation, together with the norm of jerk (J), as key characteristics for recognizing gait events using IMU data. We found that accurate gait event detection of initial contact and toe-off was achieved when the ωsagittal was within the range of -300 to -340 [deg/sec] and the J fell between 100 and 440 [G/sec]. These specific parameter ranges were effective in providing precise gait event detection. The accuracy of these measurements was approximately 25 msec, with no misclassified events detected for the specified parameter ranges.

Keywords: Rehabilitation Systems, Welfare systems, Decision Making Systems
Abstract: Home-based upper-limb rehabilitation robot is required for the hemiplegic patient to continue the rehabilitation after discharge from the hospital. Many researchers have been proposing robots dividing into body parts such as robot for arm rehabilitation and for hand rehabilitation. However, rehabilitation robots specialized for a certain body part are difficult to introduce because the patient needs to purchase a robot for each body part. To overcome this problem, we have been developing the upper-limb rehabilitation robot which can rehabilitation both the arm and the hand only by single unit. Our key idea is customizability. The robot is designed so that it is easy to assemble. We found that our robot has possibility to monitor and assist the user's rehabilitation while changing its hardware from experimental result.

Keywords: Vision Systems, Machine Learning, Decision Making Systems
Abstract: In recent years, landslides caused by excessive reclamation and natural disasters have caused many lives and economic losses. Therefore, in addition to the most fundamental method of preventing ground landslides, being able to detect signs of landslides to a certain extent is also a way to prevent loss of life and finances. Currently, satellite imagery and aerial photography are the main methods for landslide detection. However, it is sometimes impossible to confirm the ground through this method due to trees and other obstructions. Therefore, some areas are calling on local residents to record the location and size of cracks and report them to local authorities if they think there are signs of landslides. In this paper, the purpose is to develop a landslide detection system. This system combines image processing and artificial intelligence, and is paired with a multi-functional head-mounted AR display to assist professional civil engineers in identifying possible stratigraphic landslide signs. The AR display marks directions to a set destination and records signs of landslides read in real time from the camera as it moves.

Keywords: Human Interface, Machine Learning, Human Factors and Human-in-the-Loop
Abstract: The paper presented an intuitive control system using electromyography (EMG) data that is obtained from the Myo gesture control armband. The aim of this study is to enable users to control multiple devices with a single EMG device in an intuitive way. The presented system shows the ability of EMG-based gestures to control Phillips Hue that allow user to control light bulb by simple hand movements. Moreover, the authors also developed new functionalities, which is according to users’ preferences, to register gestures. These functions aim to improve the usability and enhance the naturalness of operations. Additionally, unique gestures, which less common in everyday life, is defined to reduce misrecognition when switching between different devices. Furthermore, to achieve reliable and accurate recognition of multiple gestures, the Support Vector Machine (SVM) models is considered to be a machine learning method for training processed EMG data. The experiment results demonstrate significant improvements in user experience and practical applicability in various interactive scenarios.

Keywords: Human Factors and Human-in-the-Loop, Machine Learning, Modeling and Simulating Humans
Abstract: Incorporating crowd models into robot simulators is a common practice in the field of autonomous mobile robot navigation research. Although crowd models developed in the field of crowd simulation are frequently used, there is no universal model that can be used for all scenarios. Therefore, it is essential to select an appropriate crowd model for the specific simulation environment. From this perspective, until now, we have been studying a method for selecting an appropriate crowd model for each category. This method observes the movement trajectories of pedestrians in the environment to be simulated, and classifies the environments into multiple categories based on the observation results. In this process, feature images are generated by superimposing the observation results on a time axis. The latent variables in the feature images are then extracted using an autoencoder, and the extraction results are clustered. However, the impact of overlapping time intervals (temporal granularity) and different extraction methods of latent variables, which are important in generating feature images, has not yet been clarified. In this paper, we assess their influence on category classification accuracy. Based on the obtained classification results, we also develop scenarios for selecting a crowd model for each category.

Keywords: Mechanism Design, Systems for Field Applications, Mechatronics Systems
Abstract: Wheeled mobile mechanisms are essentially unsuitable for moving over rough terrain, but the rocker bogie mechanism is known as a six-wheeled mobile mechanism with high ground adaptability that solves this problem. However, this mechanism requires a large number of degrees of freedom (DOF), at least 10 DOF, or 12 DOF in the case of an omni-directional moving function. This contains problems in terms of control system configuration, mechanism development, and energy efficiency. In this study, the number of DOFs is reduced to a maximum of 6 DOFs and a minimum of 4 DOFs by applying omni-directional wheels to the rocker bogie mechanism. First, we examined how the two main types of omni-directional wheels should be arranged in relation to the six wheels. Based on the results, an actual machine was developed for evaluation, and its effectiveness was confirmed through step over experiment on several types of stairs.

Keywords: Mechanism Design, Mechatronics Systems, Control Theory and Technology
Abstract: The purpose of this study is to develop a super directional speaker with the parabolic reflector that can arbitrarily change the width of the reflected audible area. Conventional ultrasonic speakers have not been able to control the width of the audible area. In our previous study, we proposed the method of radiating sound to a parabolic reflector and arbitrarily changing the width of the reflected sound. In this method, the sound must be emitted from the focal point of the parabola. This sound is called a point source. To generate this point source, the Curve Shaped Speaker, in which each speaker element is arranged on a circular arc, has been proposed. However, this method requires a mechanism to change the curvature of the parabolic reflector and a mechanism to change the position of the point sound source (focal point), which requires a large space. Therefore, this study proposes an ultrasonic speaker system that keeps the curvature of the parabolic reflector constant while keeping the width of the reflected sound constant. To achieve this, this paper proposes a speaker system in which the central angle of the audible area, which fans out from the focal point to the parabolic reflector, can be controlled (called a “Stepped Notch Shaped Speaker”) based on Huygens' principle. Basic experimental results confirm that the step-notched loudspeaker can concentrate sound at an arbitrary position with high sound pressure.

Keywords: Machine Learning
Abstract: This study proposes an Automatic Emoticons Insertion System that enhances text communication by estimating user emotions from speech and automatically inserting corresponding emoticons. The system uses Support Vector Machine (SVM) algorithms to classify emotions based on acoustic features extracted through the openSMILE toolkit. By optimizing SVM hyperparameters via grid search, the model achieved a significant improvement in classification accuracy, with the best performance at C=3.997 and γ=0.03071. Additionally, we introduce a visualization technique based on Plutchik’s Wheel of Emotions to represent the emotional nuances conveyed by emoticons. This approach integrates emotion recognition into digital communication, improving emotional clarity and reducing misinterpretation.

Keywords: Rehabilitation Systems, Modeling and Simulating Humans
Abstract: The objective of this paper is to identify the stresses placed on the musculoskeletal system when small dogs walk on floors with low friction coefficients. In many households, dogs are kept as pets, and small dogs in particular are often kept indoors. According to veterinary experience, it is known that small dogs kept indoors that walk with their lower limbs slipping cause a great deal of stress on the musculoskeletal system, and that the incidence of joint disease in the lower limbs of such small dogs is high. In this study, we measured the body movements of small dogs walking on slippery floors using a three-dimensional motion capture system and force plate system. Furthermore, by inputting these measurements into the musculoskeletal simulation model we have developed to simulate the musculoskeletal system of dogs, we derived the load on the musculoskeletal system of small dogs on slippery and non-slippery floors. As a result, it was identified that walking on a slippery floor causes a strong load on the lower limb muscles of small dogs.

Keywords: Autonomous Vehicle Navigation, Path Planning for Multiple Mobile Robots or Agents, Motion and Path Planning
Abstract: The objective of this paper is to develop a robot system that can transmit the video taken by the drone to the computer in real time, use YOLO to recognize the image of pedestrians, and enable the ground robot to plan the path to avoid pedestrians. External sensors, such as laser range finders (LRF) and CCD cameras, are often used on mobile robots. As we all know, these sensors are susceptible to occlusion. Therefore, even if the mobile robot avoids the pedestrian in front of it, it is likely to collide with other pedestrians that cannot be detected due to occlusion. In this study, we used a drone to shoot videos at high altitudes, transferred the videos to a computer, used YOLO on the computer to perform image recognition of pedestrians around the robot, and passed the recognition results to ROS, which enabled it to plan a path to avoid pedestrians around it. As a result, showed that the videos shot at high altitudes could effectively identify pedestrians around the robot after being recognized by YOLO.