We discuss the results of the validation and improvement of an outdoor autonomous mobile robot platform ORNE-box that we developed. On researching outdoor autonomous mobile robots, the burden of hardware development has sometimes been so heavy. It has caused problems in allocating sufficient time for the development and verification of the algorithms conducted in research. To solve this problem, we are researching and developing a robotic platform ORNE-box series. In this paper, we describe the results obtained after two years of use and the improvements made. The robot has been verified to satisfy most of the required specifications through many autonomous navigation experiments during the two years of use.

The purpose is to design a flexible object that combines soft cubes with different mechanical properties and deforms as desired. Deformation conditions for loads were determined, and design and fabrication of flexible objects were attempted. We propose an automatic design method for flexible objects. Optimal processing was performed by multi-material topology analysis and large displacement analysis, and a flexible object that met the conditions was designed. We assumed two types of soft cubes with different flexibility and tried to design them. It is necessary to model the assumed soft cube as a preliminary stage of the design. Its modeling is determined by multi-scale analysis. By combining this soft cube, the solution closest to the displacement condition is derived by Bayesian optimization. We showed the applicability of the design method proposed in this research.

In this paper, we propose a lightweight and flexible operating interface that can be implemented on inflatable mobility surfaces. It utilizes a capacitive touch sensor composed of a conductive fabric and a resistor. The sensor has three electrodes and can provide multiple inputs on a single wire. These features enables folding and fitting to curved surfaces, which are not possible with conventional interfaces.

This paper discusses the characteristics of a planar cable-driven continuum robot realizing quasi-zero stiffness. We considered a mechanism design using an elastic rod as a backbone and routing a cable in parallel to it while connecting another cable to the attachment point of the elastic rod on the end disk. This design enables the continuum robot to achieve a locally neutral static equilibrium state under constraint condition where the length of the elastic rod and cable tensions are kept at desired values. Under the above condition, static analysis was conducted to calculate the total potential energy in the surroundings of the configuration of a locally neutral static equilibrium state. The numerical calculation results showed that the continuum robot with quasi-zero stiffness exhibit properties with zero and positive stiffness or ranging from negative to positive stiffness, depending on the given configuration.

We propose a fluid soft actuator that imitates the structure of the mimosa pudica. The mimosa pudica releases Ca2+ into cells when stimulated from the outside. The leaves can be closed by using Ca2+ as a trigger to change the amount of fluid inside the cell. In addition, the leaves can be opened again by restoring the amount of fluid over time. We show that closing and opening operations can be performed by changing the amount of fluid using a soft actuator. Furthermore, we propose a valve that operates by inputting a fluid signal, and experimentally examine that it is useful for the actuator closing motion.

The McKibben pneumatic muscle (MPA) is a soft actuator with muscle-like properties that contract and exert tension upon applying compressed air and is commonly employed in musculoskeletal robots. To facilitate the implementation of tension feedback cooperative control in these robots, we have proposed that each MPA independently modulates its expansion and contraction in response to the tension exerted upon itself. In order to implement this control method within actual robots, we previously developed a compact and lightweight embedded tension sensor specifically designed for use with musculoskeletal robots. In this paper, we have enhanced the sensor’s responsiveness by adjusting the hardness of the rubber, which constitutes one of its components. The developed sensor was installed in a leg-type robot, and we confirmed its ability to measure tension via experimentation accurately. Specifically, the developed sensor could measure several hundred Newtons of tension; however, it could not measure relatively low levels of tension, i.e., several tens of Newtons.

McKibben pneumatic actuator (MPA) is a soft actuator that exerts tension. MPA control system ordinarily has a pneumatic system and electric circuit. Therefore, the system of the robot driven by MPA will become large. To solve this problem, a mechanical, pneumatic circuit has been proposed. Furthermore, MPA also has flexibility against external forces. This study examines the adaptability of pneumatic control to external factors applied to the MPA. This paper confirms that change in response when an external force is applied to MPAs connected to two oscillation circuits.

This research aims to construct a super multi-leg, super-distributed walking robot with more than 100 legs by using a design approach that distributes a large number of mechanical control devices throughout the robot’s body. The robot does not have any sensors or computers and generates periodic motions autonomously by applying a constant pressure of compressed air. In this paper, we report on the development of body modules that constitutes a highly decentralized and super-multipedal walking robot and on the development of an autonomous decentralized walking robot with a total of 18 legs by combining three modules.

This study explores the balance strategies that could underlie different leg locomotion strategies. The research focuses on the reflex circuit that generates leg movement in cats during walking, specifically the inhibitory circuit that runs from the hip flexor to the knee extensor muscle. The study demonstrates that this circuit plays a crucial role in generating leg movement that helps prevent falls.

In recent years, quadruped robots have been studied in various fields, and their usage has been increasing. While most of these robots rely on computer control to generate motion, the robot we have developed performs balance control while walking without the use of any computer, but with solely mechanism. In this paper, first, we show the structure of the robot and the gait patterns it generates. Next, we explain how the robot maintains balance while walking and its behavior, while the final section discusses the current robot’s challenges and future prospects.

Animals achieve diverse touch sensing by combining kinesthetic sensing with touch sensing on the skin’s surface. Birds’ flexible necks contain numerous receptors, making them sophisticated sensors for detecting self-posture. Therefore, the flexible neck may have a contribution to touch sensing that involves interaction with the environment. We focuses on the pecking behavior of birds and investigates the contribution of neck flexibility to touch sensing, using physical reservoir computing. This study demonstrates that the dynamics of flexible structures driven by tendons can be used for object classification. Specifically, using time series data of joint angles for learning, we achieved three-class object classification with up to 75 % accuracy. These findings highlight the possibility that flexible structures can be used for sensing as well as manipulation.

In this research, we propose an underground excavation robot that can move in the ground. This robot reproduces the movement of moles using biomimetics. However, there are few detailed documents about the movement of moles. Therefore, it is performed from the analysis of the operation of the mole. This year's research aims to elucidate the unknown movements of moles by filming their digging behavior using a transparent gel instead of soil. The objective is to develop a mole-type underground excavation robot using a water hydraulic actuator based on the data obtained from the filming.

In recent years, the progress in microfabrication has created high expectations for the development of micro-robots for medical applications. However, creating micro-robots presents several challenges, such as assembling and integrating a driving source. Additionally, it is currently difficult to control micro-robots to move as intended. In this study, we propose a rotation mechanism driven by biomolecular artificial muscle. If realized, this mechanism may enable previously ineffective artificial muscles to have a wider range of movements and translational applications. We demonstrated the feasibility of constructing this rotational mechanism and achieving a biomolecular artificial muscle-based rotational mechanism through a complex process in experiments.

This paper describes SUSTAINA-OPTM, a robot for participation in the KidSize category of the RoboCup Humanoid League. This robot was developed as a humanoid hardware platform to overcome the barrier of high hardware development costs to participate in the RoboCup Humanoid League. The concept of this platform is a ”sustainable” platform that is easy to use for new participants as well as for software developers who use many of the robots, allowing for long-term development. For long-term development, we focused on designing for increased robustness, reduced resources for repair and maintenance, and increased versatility. In addition, the system is equipped with a high-performance SoM for general-purpose software development, given the growing importance of deep learning-based recognition and other applications. The data of the developed robot is available on GitHub.

We are participating in RoboCup, a testbed for research and development of autonomous robots. We have been using a rule-based decision-making method to determine the robot's actions. However, the rule of RoboCup soccer is complex, and it is difficult to describe the decision-making rules for the robot. In this paper, we propose a method to modify rule-based decision-making without rewriting the program. We collect a dataset of decisions made by the rule-based decision-making program to train a neural network that imitates the decisions. Then we modify the dataset to change the output of the neural net as desired. The proposed method has the potential to easily modify the selected action, because the selected action can be changed by modifying the data. We describe the proposed method and verify its effectiveness.

In this study, a new small underwater robot was developed due to the aging of the previously developed small underwater robots. The developed underwater robot is sufficiently waterproof even at a depth of 11[m]. In addition, a new communication device was developed to enable the robot to switch between remotely operated vehicle and autonomous underwater vehicle. The developed communication device communicates via wired LAN, so it is less affected by the limitation of the communication cable length. As a result of verifying depth control in a test tank, it was confirmed that the system has sufficient performance even in proportional control.

This paper introduces the development of the 3D information representation method overlaying on a camera image to help a remote-controlled robot operator. This development purpose is for an operator to make it easy to understand the environment where the robot is deployed. At the same time, this method should not increase the operator’s strain, such as changing the viewpoint. Therefore, this method represents 3D information overlaying on a camera image. A camera image is most popular for remote operation’s environmental information representation, and this method aims for the operator to refer to camera images with 3D information as augmented reality. This paper describes the approach of this method, the implementation of this method for mobile robot remote operation, and the .

Radioisotope finding method is required to prevent terrorist acts using radioactive source(s) and to find lost or stolen radiation source(s). For operators’ safety, it is desirable to use an unmanned system with remote operation. Therefore, we are developing a prototype system using a 4π gamma imager and a 3D-LiDAR mounted on an unmanned vehicle for simultaneously performing gamma-ray imaging and recognition of surrounding structures. Finding of a high-intensity ¹³⁷Cs gamma point source located at 70 m square field was demonstrated experimentally based on fusion data between 4π gamma images and 3D-LiDAR-SLAM obtained by the prototype system.

As a herbicide-free weeding method, our research group are developing an autonomous running paddy weeding robot that can travel by a simple operation. The paddy weeding robot removes the weeds by churning up the soil and inhibits the growth of the weeds by blocking-off sunlight. Before now, the paddy weeding robot has detected rice seedlings with capacitive touch sensors and turned to next rice seedlings row. However, the detection accuracy to move next rice seedlings row wasn’t sufficient. In this paper, we propose a detecting system which can detect rice seedlings from images taken by a camera sensor using You Only Lock Once (YOLO). From the experimental results, we confirmed that a detection system that sorted rice seedlings by growth obtained high values in the precision and recall rates, which are evaluation indices.

As one approach to smart agriculture, we have been studying using swarm robots for weeding, burdensome part of agricultural tasks. In this paper, to perform weeding using swarm robots, we propose a method to graph the field based on path features and divide the weeding area according to the number of robots using the k-means++ method for planning robot paths. To verify the usefulness of the proposed method, experimental simulations of region segmentation with evaluation of the number of robots and the movement cost were conducted using several fields.

This paper describes a method of ego-motion estimation of a mobile robot using 2D-LiDAR in an orchard. The scan matching method is one of the methods for estimating the ego-motion using 2D-LiDAR. In an uneven environment such as an orchard, the same plane cannot always be observed, which may result in a loss of accuracy. On the other hand, the position of the tree trunk remains constant even if the plane of observation changes. Therefore, in this research, we use tree trunks as landmarks to estimate the ego-motion. In orchards, 2D-LiDAR detects not only tree trunks, but also tree branches, weeds, and the ground. We, therefore, remove them from 2D-LiDAR data to obtain the tree feature points. Then, we do matching points between previous and next frame to use congruence of triangles composed of feature points. After that, we estimate the ego-motion using the corresponding points by the least-squares method.

Agriculture relies heavily on human labor, especially in greenhouses where tasks like harvesting, spraying, and monitoring often involve long hours and repetitive work. However, the shortage of labor and increasing demand for food production has made agricultural automation a necessary solution. Mobile robot navigation is a key component of this automation, and this study presents an approach for safely navigating a mobile robot in a greenhouse using only a 2D LiDAR. A topological map and a set of behaviors allow the robot to move autonomously from one junction to another, following rows and transitioning to the next row while avoiding collisions. Our system was compared to “move base” in both simulator and testing field environments, and evaluation metrics such as path completion rate, number of interventions, and heading variance showed that the proposed system allows for smoother motion and longer autonomous travel distances.

In dairy farming, it is required to raise animals in consideration of animal welfare. In addition, it is also required to increase the amount of milk, which is the product. For these realizations, it is necessary for humans to intervene theoretically and cows to be relaxed. Therefore, we defined an index of satiety from the stomach contents of dairy cows. In addition, a behavioral model was created using a hidden semi-Markov model. By introducing an index of satiety and creating a behavior model, it was confirmed that the behavior model changed appropriately for each satiety degree. When hungry, it transitioned to eating behavior, and when full, it transitioned to non-eating behavior. We were able to create a model that matches the actual behavior.

When weeding with a robotic hand, part of a weed may remain depending on the grasping position. Therefore, it is necessary to detect a weed stem as an appropriate weeding position. A method that uses deep learning to segment weeds and detect stem positions requires pixel-wise labeling of the images in the dataset, which takes a lot of time and effort. This study proposes a method that detects a weed picking position using YOLOv7 and image processing. The proposed method identifies weeds using YOLOv7, and then detects the stem position only by combining simple image processing. This simplifies the preparation of the dataset compared to deep learning-based methods for detecting stem positions.

In this paper, we propose a one-shot learning method that enables highly accurate learning with a single or small amount of data using data amplification by GAN, and build a harvest recognition system using it. This method recognizes human harvests and trains GAN. We increase the number of harvested data using the learned GAN and learn YOLO. In experiments, it was confirmed that a discriminator can be obtained from a small amount of data, and that data amplification by GAN can improve the lack of data. As a future work, the amplified data of GAN has low accuracy, and he needs to improve the accuracy of GAN to improve the effective generator.

Our group is researching a support system for cultivating Shine-Muscat grapes. The ripeness of the grapes is determined by their color and is compared against a Shine-Muscat color chart. In this paper, we introduce a color estimation model that can determine the optimal time for harvesting by using a deep learning model to estimate the color of the grape bunch. We incorporate metric learning during model training and use the specific color space representation of images to achieve high accuracy in our color estimation model. In addition, we propose a method to incorporate an appropriate reference color into the color estimation model. Our evaluation experiment indicatesd that our proposed model correctly identified the optimal time to harvest for 85.9% of the grapes.

This research aims to develop a robot that can autonomously suppression weeds in rice paddies. Our method realizes efficient operation of the robot by estimating the position of rice seedlings in the entire paddy field from bird's-eye view images taken from diagonally above. From bird's-eye view images taken from diagonally above, various patterns of rice lines can be observed depending on the shooting position of the image, like "subjective contours". This paper proposes a method for detecting "subjective" rice lines from bird's-eye view images using deep learning and a method for estimating the location of individual rice seedlings using these rice lines.

In this study, we address to determine the cutting point for saffron picking by detecting the petal and root area of the flower using deep neural network. We detected each area by fine-tuning YOLO v3. As a result, the mAP at the end of the training was 94.59%, indicating highly accurate detection. Then, we prepared a dataset for determining cutting point based on the coordinate data of detected rectangle and performed a regression analysis using a four-layer neural network. As a result, the x-coordinate estimation has room for improvement, while the y-coordinate estimation was satisfactory.

In this paper, we research whether the reconstruction model integrated based on the 2d code is as valid as the model reconstructed from the whole image group. We investigated the difference in accuracy between the whole reconstruction model and integrate the reconstruction model in simulation. At first, we generate images assuming four linear orbits. We reconstruct three models using different three image groups, four linear orbits, two linear orbits, and another two linear orbits. We integrate two models based on 2d code. We evaluate the accuracy of the reconstruction model. We align the reconstruction model and reference model and count the points of the reference model where the distance to the nearest point of the reconstruction model is less than or equal to the threshold. We compare the whole model and the integrated model in the percentage of that point. The result shows high accuracy demonstrated with 15mm error.

This paper proposes a visual and tactile display system of a teleoperated robot for decommissioning tasks. The visual display system solves problems of blind spots of cameras and motion sickness by using the method that the arm of the robot moves a gravity compensated armstand that has a camera at its tip to the appropriate position. The tactile display system allows the operator to know the forces acting on the robot’s hand by showing the displacement of components connected to the fingertip via an elastic element using the camera of the camera stand. Several tests using a robot with the proposed systems have confirmed that the visual display system facilitates a detailed task of a teleoperated robot and the tactile display system can closely display the state of forces generated at the fingertips.

Many fuel debris still remain in the nuclear power plant where the accident occurred and decontamination work is mainly carried out by human workers. In order to reduce the exposure of workers, it is necessary to know the location and intensity of radiation sources. However, it has been difficult to locate radiation sources due to the shielding effect of structures. In this study, we used 3D object recognition to identify structures that shield radiation sources, and succeeded in compensating for the reduction in dose rate due to shielding. As a result, the estimation considering the shielding effect of the structures was realized.

Robot social implementation is expected nowadays, and in particular, for autonomous mobile robots, it is becoming necessary to be able to ”collaborate with others (other robots).” When dealing with such multi-agent systems (MAS), framework sharing has the effect of reducing the burden on designers and improving compatibility and code reuse between agents. In particular, RoboCup is a landmark project that includes the goal of accelerating development and research speed through the sharing of technology and knowledge among teams in competitions. Therefore, in order for engineering applications using MAS to progress to practical stages, a framework that supports the design, implementation, evaluation, and improvement of MAS must be available. This study aims to propose and implement a framework for the RoboCup middle-size league. ROS2 and its network were utilized for implementation, and a proposal including system configuration was made.

This paper describes a ball-throwing robot that competed in “Intelligent Robot Contest”. Author joined the contest intermittently since 2007. The robot named “SLAM-D4” is an original mechanism that was inspired by basketball. This robot equips 2DOF ball throwing mechanism that was applied snap-through buckling phenomenon. Elastic thin plates work as holder of the ball, accumulator of energy and pusher of the ball. Simple structure allows realizing the robot with off-the-shelf parts. It is also explained about ball lifting 1 DOF arm using internal spur gear line.

This research proposes a link mechanism for a robot arm that deforms during collision or contact to reduce damage to the contacting object as a countermeasure against collision safety for cooperative robots. We named the proposed mechanism “Snappable Mechanism” and conducted production and experiments. Experiments were conducted on applying an external force without moving the mechanism, and on moving the mechanism to contact an object. In the former experiment, the expected behavior of the mechanism was confirmed, but it was also confirmed that the load at which the mechanism actuated differed depending on the position of contact. The results of the latter experiment showed that when the mechanism was actuated, the load on the contacting object decreased. Future improvements could include an increase in the actuation load.

This study discusses a framework for the autonomous calibration of pipetting in a human-robot collaborative environment. Supposing that the collaborative robot works in a typical biology laboratory, in which there are unconnected devices and instruments, we developed adaptation approaches to fit the robot system for the environment. The end-effector including a commercially-available pipette and vision system was designed for pipetting and recognizing the environment. We demonstrated the autonomous calibration of pipetting with automatic robot-trajectory planning by 3D environmental information and the vision-based readout of microbalance.

In this paper, we propose an approach for realizing robotic cutting of a mouse cranium for the cranial window surgery. This task requires the skills of the minute manipulation, dealing with an individual difference, and dealing with a regional difference, which are expected abilities of autonomous robots but were not sufficiently implemented in previous studies. We propose a system recognizing if the cutting target (e.g., the mouse cranium) is pierced in the current cutting segment and avoiding cutting pierced segments to prevent the robot from damaging the mouse brain. In this system, we use vibration recognition to deal with the individual difference and apply this to each cutting segment to deal with the regional difference. We confirmed this system work against the simulated cranial window surgery using chicken eggs.

Sit-to-stand (STS) is one of the most demanding activities of daily living, especially for older adults. In this study, we propose a pneumatic soft exosuit for sit to stand assist. The proposed exosuit is powered by variable stiffness pneumatic muscle (VSPM), which can freely deform while deflated, but straightens when inflated. This allows the VSPM to be directly provide power assistance to the human body, while also conforming to the differences between individuals. Four VSPMs were attached to the exosuit, which will allow the exosuit to generate enough moment required for STS in both the knee and the hip joint. Future work includes measuring the moment output, and validation of the effectiveness of the developed suit in human trials.

In the manufacturing and transportation industries, lifting activities cause back pain. As a countermeasure, lumbar support suits have been developed. Among lumbar support suits, exoskeleton-type suits provide strong support but are heavy and difficult to move around in. On the other hand, endoskeleton-type products are made of soft materials, so they adhere closely to the human body and are easy to move around in. However, they have the problem of weak assistive force. Therefore, in this study, we developed a product with an ingenious design to achieve both ease of movement and strong assistive force, while keeping the structure simple to keep costs down.

When constructing marine infrastructures, the foundation stones that have been dropped into the sea are leveled manually by divers. At this time, a large strain is placed on the diver’s body. Therefore, in this study, an assist suit for elbow joints that can be used underwater was developed to reduce the burden on the diver’s arms. In order to evaluate assist effects of the developed assist suit, experiments of measurement of electromyogram and expired gas analysis were carried out. From the experimental results, reduction of the physical burden on the subjects was verified for the rolling work of the heavy object.

Assistive suits are being researched and developed, and some of them are already in practical use. However, many assistive suits in practical use often block with complex human movement because the stiffness of the motors and large frames that make up the suit limit the degree-of-freedom of the user’s joint. In previous research, we have attempted to solve this problem by developing an assistive suit with a flexible drive unit that combines soft actuators, McKibben muscles, and soft clothing. To improve the performance of this suit while maintaining its flexible characteristics, we also conducted studies using planar frames in part. In this study, we made a prototype arm support suit combining McKibben muscles and planar frames arranged on the body surface to support elbow flexion and measured the movable joint angles.

This paper describes the experiments and the research into the effects of an interface accompanied by a physical sensation. It proposes a concise, intuitive, and comfortable operating experience for the interface that changes the function of a lighting fixture by accompanying it with the physical sensation of a Membranophone, a membrane-sounding instrument such as a drum. The specific research and the experiments include: examining the relationship between people providing intuitive operations, examining methods for detecting placement and strikes, and examining methods for emphasizing the physical sensations caused by the effects of the light accompanying the operations. Through this research, we hope that the interface will improve the user's experience and contribute to the development of interactive attractiveness in electronic devices.

This study aims to construct a performance support system that enables anyone to participate in playing musical instruments by using mechatronics and robotics technologies, and to share the musical experience with fellow players. The system under construction was developed through a design thinking approach, assuming hearing-impaired persons and their supporters as users. This paper reports on the process and the system concept.

Students are engaged in problem-based learning for special needs education. This paper proposes a visual support system for playing instruments in order to share musical experiences with participants. Prototype system has been developed for a traditional Indonesian music ensemble called as Angklung. The system can read a pre-created music file and create a musical score. It can color-code the musical scale to be played and present the timing of percussion. The tempo of the music can also be changed. Through the problem solving in the field, the students are able to broaden their perspectives on international and special needs.

Starting with home appliances, Hitachi aims to provide new value to customers by expanding solutions to household chores, home life, and urban areas. This paper reports on the development and use case verification of a mobile air purifier, which is one of the ideas extracted from customer surveys.

In this study, we develop a wearable finger that can be operated intuitively to support daily lives of people with temporary one-hand disabilities. The finger mechanism is based on a four-node linkage mechanism. The device comprises two fingers opposite to each other and can perform not only simply grasping but also pinching an object. The fingers are driven by a single servo motor in order to avoid the problems related to weight and control complexity. The device employs EMG signal-based switching control, allowing the user to operate these robotic fingers in a similar manner to his/her ordinary physical movements. Experiments using the device and control method verify that it can grasp weights of up to 1200 g and 17 different objects. Furthermore, the device was able to perform simple daily activities involving the grasped object in place of a disabled hand.

The hip joint exoskeleton causes pain at the position of application due to misalignment of the hip joint exoskeleton. To solve this problem, we applied a parallel link mechanism. We designed and fabricated a hip joint exoskeleton using four parallel link mechanisms so that the range of motion of the hip joint exoskeleton can cover the range of motion of walking. A force sensor was attached to the fabricated hip exoskeleton by shifting its position on the evaluation device. The evaluation was performed by measuring the force and -motor moment generated at the position of the thigh. The measured values showed that a large force was generated when the hip joint approached a singular posture.]

In the aging society, maintaining gait function and motion is important for a hemiplegic patient who left the hospital. In this study, we propose a human-powered and wearable mechanism that assists the hip joint extension of the paretic side and the flexion of the non-paretic side by use of the user’s residual function, especially non-paretic hip joint motion. In this paper, a prototype design was shown and a link model of the lower limbs and the proposed mechanism were explained. With the paretic gait motion data, the accumulation torque and assist torque was estimated by use of the proposed link model. In the proposed mechanism, the estimated assist torque was 2.7Nm for the paretic side and 3.1Nm for the non-paretic side. In future work, we will focus on full body motion and redesign the proposed mechanism in detail to manufacture the prototype and evaluate the function of gait assistance for hemiplegic patients.

In modern society, where the 100-year lifespan is approaching, not only is the musculoskeletal system debilitated due to the super-aging problem, but also due to the increase in the number of people with locomotive syndrome, everyone is facing a crisis in which they cannot maintain their health due to lack of exercise. In order to solve these problems, researches on wearable robots that aim to assist and extend human body movements by wearing them are being actively conducted. Wearable robots often use surface electromyogram as human motion intention. In this study, we developed a flexible sensor that can measure displacement mechanomyogram, which correlates with surface electromyogram, in order to estimate human motion intention more easily, and verified its usefulness.

In recent years, with the development of robotic technology, prosthetic hands with sophisticated functions have been actively researched. On the other hands, in terms of size, number of parts, and customization, low-cost, functional pediatric prosthetic hand with compliant wrist joint hasn’t been developed yet. In this study, we developed a 2-DOF wrist joint for pediatric prosthetic hand using a compliant mechanism, constructed its design methodology, and conducted an evaluation experiment using a prototype. The proposed structure enables circular motion by combining four movements: flexion, extension, adduction, abduction, and succeeded in acquiring movability similar to the human skeletal structure by giving anisotropy to the arrangement of the elastic hinges. It can be manufactured monolithically, which is advantageous for mass production and assembly.