Due to the declining birthrate and aging population, the convenience store industry is facing a serious labor shortage. Therefore, there is a strong need to automate the display and disposal of products in convenience stores. In our previous research, we have developed a display and disposal robot to reduce the workload of employees. However, the developed display robot has limitations in moving in the Z direction. Therefore, we improved the system that extended the range of motion in the Z-direction by extending the length of the link and using the elevation mechanism. As a result of improvement, the range of motion of telescopic mechanism has been extended to 240 mm in the Z-direction, and combined with the lifting mechanism, the robot hand can move above the products on the shelf.

Since security guards are in short supply, it is necessary to introduce security robots to provide efficient security. Security robots are required to have autonomous movement and teleoperation functions depending on the situation. However, the field of view of camera images used for teleoperation is limited. In our laboratory, we have developed the system to overlay the environmental information on the camera image to assist in understanding the environmental information during teleoperation. In this research, we evaluated the operability of the system during teleoperation through experiments. As a result, it was shown that the extension line display contributes to the improvement of work efficiency and the LRF distance data display contributes to the reduction of workload.

The purpose of this research is embed a QR code image into any image by means of deep learning. In this research, I create Encoder model that embeds QR code image into an image and Decoder model that recovers the information of the embedded QR code image referring to the AutoEncoder model. The loss parameters of the two models are shared and optimized.

By introducing embedded technology into ROS 2, it is expected that real-time performance will be improved and power consumption will be reduced. The purpose of this research is to compare various runtime environments of ROS 2 nodes for embedded devices. We take up mROS 2 and micro-ROS as targets, and evaluate them in terms of communication performance and memory size. We discuss the advantages of each environment while comparing them not only quantitatively but also qualitatively.

Robots that can coexist with humans are needed in manufacturing and service fields, but there are issues in ensuring safety. SafeML, which is a model-based approach to develop safe robots, has been proposed in the past. This time, we extend the specification to such human-machine cooperation and introduce a model example.

In the ROS, which is middleware for developing robot control systems, the programs that realize the robot functions are treated as nodes. In ROS 2, the latest version of ROS, communication between nodes is realized by DDS, and several vendors provide DDS implementations with different characteristics. However, in the current ROS 2, only one DDS implementation can be selected per system. Therefore, it is difficult for all nodes to communicate with each other using the optimal DDS implementation due to characteristics such as data size and communication range. In this research, we propose a mechanism called“DDS Binder”that switches the appropriate DDS implementation for each communication path between nodes. We realized the switching of DDS implementations under the constraint that only one DDS implementation can be used simultaneously in one node, and the latency measurement results showed that the inter-node communication was optimized.

Risk assessment is one of the important processes in the social implementation of robots. In risk assessment and safety design of systems with various stakeholders, modeling and visualization of related elements are important not only for designers but also for users. The use of SafeML in the process was examined and its extension for the purpose was proposed for the missing elements.

The ELTRES developed by Sony is an IoT device capable of communicating over 100 km despite its low power. ELTRES has a built-in GPS to synchronize data between transmitter and receiver in order to achieve long-distance communication at low power. ELTRES has a stand-alone function to transmit location information as 128-bit data, but it can also send 128-bit data from outside. In this study, we developed a method of compressing the location information obtained from ELTRES at regular time intervals and transmitting 12 to 30 points of location information every minute. In this paper, we introduce the developed system and report the results of measurements using the system.

An international standard called the R&K method has been established as a method to identify sleep stages. This method is not suitable for daily monitoring because it requires a large number of contact sensors and a doctor’s judgment. In this study, we are developing a system for non-contact classification of sleep stages using commodity Wi-Fi devices. In this paper, we describe the elements of the system, such as “Acquisition of biometric information” and “Acquisition of body movement information”.

This study aims to develop a cargo handling robot that can be put to daily use, considering factors such as high throughput, high mobility, smoothness, simplicity and low cost of construction. An inverted pendulum-wheeled mobile robot with a transformable body is proposed, which adapts to different road surfaces by switching between two and four wheels traveling modes. The four-wheeled mode state can traverse obstacles, where as the two-wheeled mode transforms into an inverted pendulum robot with higher maneuverability. This enables turning in tight spaces and moving smoothly and quickly in flat terrains.

This paper describes the construction of a simulator for conducting large-scale robot experiments. In recent years, the applications of mobile robots have become more and more diverse. Many mobile robots will be widespread for general consumers in our living environments. This study considers to see large-scale cooperative systems of robots and systems for efficiently sharing huge amounts of data. A large-scale robot simulator using Unity, ROS, Docker and Ethereum for data sharing is developed. Especially, grid map sharing and navigation in multiple mobile robot network is introduced as an example.

The purpose of this study is to develop a system that allows any user to easily connect to the ROS-based robot from a smartphone application at any time and remotely execute multiple ROS commands of the robot according to the user’s purpose. In addition, as an operational example, a remotely controlled mobile robot was developed. This robot system can be operated remotely for a long period of time by low level control commands and an automatic docking command to the charging stand. Also, the developed smart phone application can be applied to any ROS-based mobile robots.

In this paper, we propose a method to reduce the force applied by humans to a robot for climbing stairs in cooperation with a robot. In the former study, we were able to reduce the amount of work, but not the maximum value of force. Therefore, we reviewed the principle of stair climbing and considered the contact area between the wheel and the floor or wall as a surface contact. In addition, we decided to maximize the torque by position control instead of force control. We thought that the human would exert the optimum force to lift the robot up the stairs, and conducted simulations. After that, we confirmed the effectiveness of this method in actual experiments. The results show that the peak value of the force that humans should apply to the robot can be reduced by position control.

In this paper, we propose a method to reduce the computational complexity of Look-up Table(LUT) by utilizing the unique features of eccentric magnetic encoder. In the previous study, eccentric rotation was used to achieve higher resolution without compromising the absolute angle. However, the number of sensors is reduced in relation to the number of magnetic poles of the magnet, a large angle error will occur. This problem can be solved by using LUT. But LUT require a large number of calculations and are heavy to process. To solve this problem, we propose to reduce the capacity of the LUT. Simulation results show that the proposed method achieves the same level of accuracy as the conventional method while reducing the number of data.

In this paper, we developed a Mobile Module (MoMo) 5.3 equipped with a wireless communication function and an application that allows the user to remote control MoMo, and constructed a system that allows remote control MoMo.In this system, the path for MoMo is automatically determined by the user inputting the obstacle position, start position, and target position.It is possible to move MoMo to any location.In addition, it is possible to move two MoMo on the simulation.In the experiment, it was confirmed that the system operates correctly through a simulation consisting of multiple scenarios and one actual machine experiment.

In this study, nonlinear model predictive control is applied to a crawler-type stair-climbing robot. Currently, the robot in this study is controlled based on a linear model that linearizes the target system. Therefore, it is not possible to control the swinging up of the front wheels. To overcome these shortcomings, we applied nonlinear model predictive control.

With the declining birthrate, aging society, and declining population in Japan, robots have been attracting attention as a means to carry out logistics. However, there is a risk of robot failure due to the impact of falling down stairs when the robot runs in the city. In this study, we propose a control method to mitigate the impact of falling stairs by using a robot with a tristar wheel.

In recent years, with the declining birthrate and the aging society, the shrinking economy, the declining labor force, and the shrinking social security system have become problems. Therefore, the introduction of vehicles to assist people in their work and daily life has been promoted in the real world. The purpose of this study is to make the vehicle easier for people to use by controlling the speed of an inverted two-wheeled vehicle according to the pulling force of a person and making the lifting force of a person lighter than the weight of the vehicle. Simulations on a level ground and on a staircase are performed and evaluated.

The information required for location estimation using RF signals is AoA and ToF. In Wi-Fi location estimation, it is possible to obtain AoA by applying MUSIC algorithm, but it is difficult to obtain accurate ToF. In this study, we focused on the Fresnel zone. The amplitude of the signal when a person is in the Fresnel zone is compared with the Fresnel diffraction model to estimate the number of the Fresnel zone in which the person is. By combining the identification of the Fresnel zone with the AoA estimation, we can estimate the location information.

In this research, we propose a structure to reduce the number of sensors in magnetic force difference type magnetic encoders. In the previous research, the problem of not being able to calculate the absolute angle due to the multipolarity of the magnet was solved by eccentric rotation, which enabled high resolution and calculation of absolute angle. However, when the number of sensors is less than the number of poles of the magnet, a large error occurs in a specific section. Therefore, we devised a method to calculate angles with high accuracy using fewer sensors than the number of poles by adjusting the magnetic force of the multi-pole magnet using the magnetic force difference for each pole.

To understand human-centric images and videos, detecting human-object interactions is an important probrem. Motion recognition based on skeleton information has been studied, but it is not possible to accurately estimate target object using only skeleton information. Therefore, we propose a method for detecting human-object interactions using the skeleton information in conjuction with the object recognition information. We define object-specific interaction by this method. Our hypothesis is that by training human-object interactions using the same recognition method as a human, sufficient recognition accuracy can be obtained even with a small dataset. Since we use only coordinates for training, robust recognition is possible in various situations.

In this research, we propose a robot hand using a closed link mechanism. The proposed robot hand (HSCR-HA) is used for communication with the deaf. The HSCR-HA has the closed link mechanism, which enables bending and stretching of fingers by wires and springs. Response characteristics of joint angles of fingers were experimentally evaluated. From the experimental results, The each settling time of each joint angles of index finger was 247 ms for the first joint (DIP joint), 238 ms for the second joint (PIP joint), and 175 ms for the third joint (MP joint). In addition, the effect of the hand direction is negligible for use as communication with the deaf.

When humanoid robots can walk on mud, they will be useful in the event of a disaster, and the range of human substitutions will expand. A walking experiment was conducted on the mud as a foothold, and the adhesive strength of the sole was evaluated.

In this paper, we attempt to design and develop a robot that mimics the human shoulder joint to achieve morphological and functional similarity. In order to successfully realize the internal/external rotation and abduction/adduction motion of the human shoulder joint, we propose a novel mechanism incorporating a hypoid gear that enables geometrical structure to be orthogonal relationship. First, the human shoulder joint is described to determine the arrangement of the degrees of freedom of the shoulder joint mechanism. We can note that the proposed mechanism with hypoid gear has superior characteristics such as high gear ratio, compactness, and geometrical advantage. Next, the designed and developed shoulder joint mechanism is actually introduced in detail. We also compare the shoulder joint mechanism with the human body dimensional data. Finally, the frequency characteristics of the robot are investigated for performance evaluation.

The arch structure of human foot absorbs impact and assists push-off movements during walking. The objective of this study is to introduce arch structures and toe joints into a biped robot, and verify the effects and devise walk control methods by dynamic simulation. We simulated the upright state and the start of walking, using the Choreonoid. From the results, it was confirmed that the arch structure improved the impact absorption and stability in the anteroposterior direction. In addition, it was confirmed that the arch structure could be expected to smooth the load transfer between the supporting legs during the step change.

Currently, humanoids have fewer leg movements than humans. In order to realize large leg movements like humans, the foot is important because it is the part of the body that makes contact with the ground. In this study, we focus on the outer edge of the foot, which has not been paid attention to so, and develop a robot foot with an outer edge measurement structure equipped with one-axis force sensors on the bottom and sides. Using this, a large rotation angle was achieved by controlling the frictional force, especially for the rotational motion in the chair seating state, which is a motion that requires the use of the outer edge of the foot.

There is a way to utilize the human-like proportions of humanoid robots to mimic human behavior. In general, motion capture is used, but in this case, although it can capture the posture of the body links, it cannot capture detailed information about the feet. In this study, we developed a device to measure the contact pressure distribution at the outer edge of the foot and proposed an index called SS-COP which simply reflects the contact on the curved surface of the foot. In the experiment, we performed ground state mimicry and pedal operation on the ground to evaluate the posture change of the human foot and the robot foot.

In this study, we proposed a walking system for humanoids that considers environmental terrain and the robot balance. While walking, the robot modifies its landing position in real time according to the robot state. When walking on uneven terrain, the robot must consider the environment at the same time and land on an appropriate foothold. In the proposed system, the environmental point cloud is transformed online into a 2D polygon representing the steppable region. After that, the robot modifies the landing position considering the environmental terrain by geometric computation with high period using this polygon. Through experiments using an actual robot, we verified that the proposed system is effective for walking on uneven terrain.

When a humanoid robot manipulates an object, it is necessary to control the whole-body motion by fully considering the manipulation force corresponding to the object and the reaction force the robot receives during the motion. When the required manipulation force is unknown, there is no choice but to passively control the whole-body. However, when the manipulation force changes rapidly, the robot cannot fully respond and falls down. In order to solve this problem, we have developed a system in which the required force trajectory is acquired based on the previous unsuccessful trial and used in the next trial to achieve a successful motion. As an example of manipulation of an heavy unstable object, we confirmed that the proposed system enables a robot to push up the front wheels a wheelchair by multiple trials, and verified the effectiveness of it.

In this study, we modified a commercially available small bipedal robot to be able to walk on the Raspberry Pi 4. First, to prepare the experimental environment, we modified the biped robot and made a stand. Then, we checked the operation of the motors. After confirming the operation of the motor, we created a program to make the robot do the walking motion. Then, using the program, we created a walking motion and conducted a walking experiment. Now that we have confirmed that the biped robot can walk, we would like to conduct research to apply the information obtained from 3D spatial recognition to biped walking by installing a depth imager on the robot.

Although continuous hand rehabilitation even at home is needed for hemiplegic patient to use him/her hand in daily life, a patient cannot be intervened by therapist compared with in clinic. In this study, we have been trying to solve these problems by development of hand rehabilitation device for self-rehabilitation at home. The device instructs a patient to lift index finger so that a patient can regain an independent finger movement, and assists lifting when a patient cannot lift. However, detailed implementation which the device makes smaller while meeting the requirements of assist is unclear. Therefore, we make airbag connected with small air pump for actuator and it is shown in this paper. As a result of experiment which checks whether it has enough ability to assist lifting, we can find it produces joint displacement by appropriate assist force.

Walking on a treadmill have been used clinically as an effective rehabilitation method for elderly people who have difficulty walking in a standing position. Rehabilitation of the lower limb is primarily made with moving the belt of the treadmill. However, walking is a motion using the lower limbs while balancing a rhythmic movement of the upper body. In this paper，we develop a new walking rehabilitation system by introduce a rotary chest support pad into a treadmill system. Walking ability of the elderly people can be expected to recover quickly using the proposed system.

The authors are developing a paralyzed hand assist system that combines functional electrical stimulation and an exoskeleton-type assist mechanism for the purpose of reconstructing hand function and assisting in daily life for upper limb paraplegics. The exoskeleton type assist mechanism of this system is designed to assist the adduction and abduction of fingers. In this paper, we describe the development of an assist mechanism using a cam that assists the adduction and abduction of the index, ring, and little fingers. The servo motor used in the assist mechanism was limited to an appropriate torque by inserting a resistor. As a result of mounting the assist mechanism on an actual hand, it was confirmed that the assist mechanism was capable of assisting the adduction and abduction of the index, ring, and little fingers, although there was some slippage.

We have developed a passive motion device for pronation and supination of the forearm. In this paper, we describe the design and prototype of the passive motion device. Pneumatic soft actuators were used as the driving source of the device. We investigated a structure of the device that moves pronation and supination of the forearm using the pneumatic soft actuators. Based on the exploratory examination, we designed the equipment and also manufactured it. In the test, the subject was measured for the passive range of motion. As a result, the prototype performed passive movements of the pronation and supination for the forearm.

The voluntary motor skills of stroke patients are significantly reduced due to the sequelae. Even in the gait training stage, there are many cases in which voluntary gait is not sufficient, leading to abnormal gait. Therefore, it is thought that voluntary extension rehabilitation after voluntary onset is lacking. However, no engineering support method has been proposed for the entire lower extremity in the acute stage, with a device to improve voluntariness. Therefore, the purpose of this study was to develop an assist system that agnostically improves the extrusion and traction force of the lower limbs using our bipedal coordination device.

Unilateral spatial neglect (USN) is defined as impaired attention to stimuli on one side, which can exist for near and far spaces combined or independently. Thus, quantifying both near- and far-space neglect is crucial. This study aims to propose an index to quantify the near/far spatial neglect ratio to describe neglect symptoms’ characteristics in each patient using virtual reality (VR). Objects were presented in a VR space and the positional data of them recognized by the patients was recorded. The near/far neglect ratio (NF_Ratio) which was calculated using proposed equation suggested that four patients tended to have greater far-space neglect, and one tended to have greater near-space neglect. Moreover, NF_Ratio was shown to vary according to height. This is the first study to propose VR system to quantify the near and far spatial neglect. However, further study is needed to assess its reliability and validity and describe its clinical usability.

Living in a bed and a wheelchair on daily basis reduces muscle strength, causing decreases in various mental and physical functions. Such functional deteriorations are collectively called the disuse syndrome. In order to overcome the disuse syndrome, to move joints and muscles is very important. Ankle contracture is one of the disuse syndromes of the lower limbs. CPM devices can often prevent/improve the symptom. But the distributed devices cannot dorsiflex ankles alternately with strong power. We then develop the simple CPM device which has small size, high power, and ankles' interlocking system. The structure and evaluation are described in this paper.

Hand movement is an essential motor function in daily life, and maintaining the motor function of the hand is important to prevent a decline in quality of life. Hand exercises are effective in preventing dementia, promoting health, and improving motor functions, and are actively incorporated into rehabilitation. However, the accelerating aging of the population has led to an increase in the number of patients and a shortage of therapists, so we need a system that can a patient evaluates the motor function of his own by oneself. In this paper, we aim to realize a system that enables quantitative evaluation of hand movement functions by games using a motion sensor, Leapmotion, and a recu rrent neural network.

We have developed a device called "PDFin" to simulate manual therapy to reduce spasticity in hemiplegic stroke patients, and have achieved a certain effect. In the future, it is also necessary to develop a protocol for rehabilitation using PDFin. Therefore, in this study, we developed two sensors to measure the finger joint angle and the force to press the hand against the device while using PDFin. Furthermore, by applying the PDFin to six healthy subjects, we verified the effectiveness of increasing the range of motion of finger joints before and after use. We also verified the safety of PDFin in terms of not causing any health problems.

The authors are developing a glider-type underwater robot for environmental survey. In this paper, we propose a method to estimate the gliding ratio and moving velocity of the underwater glider under steady state conditions, based on the force balance and the measurement results of the fluid force acting on a small model of an underwater glider. The estimated results were compared with the results of an underwater gliding experiment using a small model underwater glider. As a result of detailed investigation, it was found that slightly 0.5[mm] error was involved in the initially measured center of gravity position. By correcting for this error, the gliding ratio and moving velocity can be accurately estimated.

High frequently measurement techniques of sea surface conditions are important for ocean science research, such as the construction of climate change models. We are developing a sea surface conditions measurement system using small buoys with multi-IMU sensors. The proposed system aims to improve accuracy with multiple IMU sensors and utilize wave periodicity to reduce integration errors. In this paper, we report the wave direction estimation method using a small buoy with two IMU sensors. The wave direction is estimated using the buoy motion compensation method which we are proposing. The wave direction estimated by the proposed method is compared with the motion capture data. Experimental results show that wave direction can be estimated by the proposed method.

The purpose of this study is to evaluate the turning and diving performance of a disk-shaped underwater robot (Cakram) with a traveling wave propulsion mechanism. The Cakram can generate traveling waves when there is a phase difference of the same frequency between the fin strips. The fin pushes the water by the traveling wave. The reaction force received from the water is used as the driving force. The turning and diving performance of Cakram is evaluated for various parameters such as amplitude difference and offset angle. The turning and diving characteristic were evaluated by the speed, trajectory and diving angle of the Cakram. The results show that the maximum angular velocity becomes 0.24 rad/s by increasing the amplitude difference between left and right fins. The results also indicated that the diving angle can be controlled using the offset angle.

There are underwater vehicles equipped with pan-tilt cameras and lights for the purpose of environment recognition. However, since most lights are fixed in directoin, the environment cannot be recognized depending on the shooting direction of the pan-tilt camera. A few devices have been developed to pan-tilt the camera and lights with individually waterproof drives, but they can only be used up to a depth of about 100m. Here, we propose an underwater pan-tilt camera device that also pan-tilts the light.

In this paper, we investigate the basic characteristics of PTFE (polytetrafluoroethylene) bellows for the buoyancy control module mounted on the environmental surveying underwater robot. First, the pressure tolerance of the bellows was measured, and it was confirmed that the bellows could be used as a buoyancy controller up to a water depth of about 50[m]. Then, we measured the spring constant of the bellows, which is necessary to calculate the energy efficiency when controlling the buoyancy force. In addition, the displacement of the bellows was simulated. The results showed consistency between the simulated displacement and the displacement calculated from the measured spring constant. Therefore, it is thought that the displacement of the bellows can be estimated by simulation in the future.

The typical control of a cable-driven parallel system is based on the assumption that the cable restraining the suspended object is straight. However, in an underwater cable-driven parallel system such as the one considered in this study, the suspended object and the cable are affected by the fluid. This causes deflection and vibration of the cable, which makes it difficult to control the position and attitude of the suspended object. In this paper, we propose a mathematical model of an underwater cable-driven parallel system consisting of multiple on-water robots that take into account the dynamics of the wires affected by the fluid.

This paper designs an underwater map construction algorithm for ROV. The algorithm takes ROS as the framework, uses imaging sonar and IMU sensor, and completes path display and map mosaic based on OpenCV. This algorithm obtains the real-time pose and moving path through IMU, and stitches the sonar imaging at each interval on the moving path, so as to obtain the map of the whole region. In the aspect of mileage calculation, this algorithm calculates the difference based on the distance measured by sonar and combines it with IMU angle. Compared with the direct integration calculation using inertial navigation, the error is smaller and the accuracy of the map is improved.

We have focused on the high maneuverability of fish such as rapid turning and acceleration, and have studied fish-like robots using snap-through buckling. In our previous research, we developed a fish robot with a bendable body consisting of a multi-link structure. The robot achieved a swimming motion similar to that of a fish that bends its body. However, the robot did not achieve the rapid turning motion like a fish due to its low swimming speed. In this paper, the material of the elastic body is changed to improve the swimming speed, and the effect of the change is verified by actual experiments. After developing a modified fish robot for verification, swimming experiments were conducted using two types of elastic materials. From the results of the experiment, it was confirmed that the swimming speed was faster using the hard metal plate than the polypropylene plate.

This study considers automatic scallop harvest using the drone with the robotic arm as the automation of the scallop fishery. The method needs the dedicated gripper attaching to the robot arm. This study judges the adsorption gripper as the dedicated gripper. When gripping an object with an adsorption gripper, the negative pressure inside the gripper during grasping is important. Negative pressure is generated by enlarging the internal space of the gripper. Therefore, this study investigated the relationship between the increase in volume of the enclosed space and the generated negative pressure. From the investigation result, the negative pressure is sufficient to grasp the scallop.

In order to conserve the coastal environment, conservation of eelgrass beds is progressed by volunteer group. Since the seabed needs to be softened for the growth of eelgrass, seabed cultivation is planned. In this article, we have been developing the small underwater cultivating robot named “EXPLOWER”. EXPLOWER is made by electrifying a cultivator on land. The developed robot was tested in sand box as land operation test. As results of the test, the robot is able to cultivate soft ground as sandbox, but cultivation claws were digging to downward. Therefore, the robot seems to have enough power to cultivate seabed, but arrangement of propulsion mechanism is need to reconsider.

We have focused on the high maneuverability of fish such as rapid turning and acceleration, and have studied fish-like robots using snap-through buckling. In our previous study, we developed a waterproof fish robot. However, it was difficult to make the robot’s shape streamlined like a fish because the waterproof box is box-shaped. In this paper, we develop a waterproof box in the shape of a fish head in order to develop a robot with a more fish-like shape. By placing the control and drive units in the head, the degree of freedom in the design of the torso is increased. In addition, we aim to improve the frequency of buckling and swimming speed by using a higher power motor. As a result of swimming experiments, we succeeded in improving the frequency of buckling of the jumping motion. On the other hand, it was confirmed that the swimming speed decreased when the frequency was too high.