We investigated the relationship between flight performance and size of a X-wing type flapping UAV. The flapping UAV equipped with a double motor drive system to control the flapping frequencies of left and right wings independently. UAVs of various sizes with different fuselage lengths, wing spans, and tail lengths were fabricated, and their lift and mobility were measured. The gearbox structure driving the flapping wings was modified to fit a variety of aircraft sizes. The lift force decreased as the aircraft size decreases. Furthermore, the lift force was largely dependent on the wing span, and there is an optimal span. On the other hand, mobility was found to increase with decreasing aircraft size.

This paper proposes a hammer-sounding wall tile inspection system using a two-wheeled multicopter. The proposed system consists of a two-wheeled multicopter equipped with a tapping hammer and a machine winding the wire connected to the multicopter. The winding machine is installed on the rooftop to hang and move the multicopter vertically on the wall. To show the effectiveness of the proposed system, we first collect hammer-sounding inspection data using the proposed system. The collected audio data is analyzed in the frequency domain to detect the delaminated tiles on the inspected walls. We finally evaluate the accuracy of the inspections using the proposed system and compare the accuracy with an inspection by an inspector.

This paper proposes a method for estimating the lateral position of a tethered multicopter and a method for suppressing wire tension. First, we describe the modeling of a tethered multicopter and catenary theory. Then we show how to estimate the lateral position and tension of the wire tip using catenary theory. Next, we design a position tracking controller for a tethered multicopter. This controller can be used to suppress the effect of the wire tension on the airframe. Finally, to examine the usefulness of the proposed method, we present the results of three types of experiments conducted on actual aircraft: the first is an experiment to verify the accuracy of lateral position estimation, the second is a flight experiment using lateral position estimates, and the third is a flight experiment using tension feedback.

The thrust of an axial-flow impeller may become oscillatory at high speed due to disturbances in the thrust direction and the natural vibration of the propeller. Therefore, a reference governor is designed based on the critical speed identified from the force sensor. The control target in this paper is making angular velocity response measured from the angle sensor track angular velocity command with suppressing an impeller vibration in the thrust direction near critical speed by compensating reference according to angular velocity response. Proposed angular velocity controller suppressed the vibration in the thrust direction by avoiding the operation near critical speed by in the experiment. Hence, the superiority of the proposed method is demonstrated in the experiment.

In this paper, we present an aircraft autopilot system focusing on a landing flight. The landing flight is composed of approaching and flaring scenarios in which different controls are required. We have presented two flight controllers based on deep neural networks. For the flight controllers, the current flight was estimated from cockpit image inputs through a proposed scenario classifier based on CNN. Due to a so-called chattering problem, however, the two flight controllers were repeatedly used. Since the speed and angle are determined by the controllers, the aircraft might fail in the landing flight. For this challenge, we propose to apply an LSTM block to the scenario classifier. The CNN with LSTM enables the aircraft to estimate the flight scenario taking time-series variation of image inputs and scenario outputs into account. In the simulation experiments, we show that the aircraft succeeds in the landing flight by switching the controllers correctly through the time-series scenario classification.

This study reports on an underwater multicopter that can take off from land and dive into the water to enable long-duration underwater surveys at arbitrary locations. We believe that the use of an underwater multicopter is an effective method to reduce manpower and time when inspecting and surveying underwater infrastructural facilities. Ultimately, it is envisioned that the aircraft will be able to estimate its own position and perform underwater autopilot operation.

In this presentation, we will propose a communication method between the aircraft and the ground underwater using buoys and Raspberry Pi, develop a positioning system using GPS and sonar, and report on how to achieve more precise flight and underwater maneuvering.

We are developing a tilting-frame multi-rotor UAV that can change the pitch angle of the airframe while keeping the UAV’s propeller horizontal, working in high and dangerous places that are out of reach. The tilting frame multi rotor UAV is fitted with a work arm for inspecting high, sloping, and curved wall surfaces. In this study, we developed a system to measure the angle of the wall surface to the work arm using four TOF distance sensors mounted on the connection between the work arm and the fuselage. In some cases, the angle of the wall could be estimated accurately, but in other cases there was a large error of around 10 degree. We also evaluated the distance measurement characteristics of the TOF sensor using targets made of different materials and with different orientations.

This paper presents a simple and practical approach for generating a complex path vector field to guide and control flying robots. The proposed approach is based on the idea of generating guidance vectors of flying robots from a given path. A key feature is to easily and intuitively generate guidance vectors that effectively avoid singularities and obstacles on the path. The effectiveness of the approach is demonstrated using a powered paraglider simulator that accurately represents complex 3D dynamics.

This paper presents a practical control approach for complex path tracking of a Powered Paraglider (PPG) and provides flight experimental results. First, a path tracking error system of the PPG is represented using the Serret-Frenet frame. Next, we derive linear matrix inequality (LMI) conditions that stabilize the error system. A guaranteed cost controller considering an input saturation is designed by solving the LMI conditions. The effectiveness of the proposed approach is demonstrated through three different path tracking experiments conducted in Hokkaido.

The aim of this research is to explore an autonomous control architecture for UAS, allowing for HOOTL operation, which is currently used for UAS operation. This study aims to investigate an autonomous control architecture for UAS to enable HOOTL operation, which is the current UAS operation method, from operator operation to HITL operation, through human on the loop (HOTL), which includes human approval. In particular, an urgent issue in UAS is the modeling of control and operation in collision avoidance and traffic management. Here, we propose constructing a reference architecture in collision avoidance and traffic management by considering the inclusion architecture and the input/output of functions. As a result, we illustrate the coupled relationship of three functions and functional components connected to eight granularities.

This report describes the development of an assessment tool for prosthesis terminal device control. The light-weight wireless tool is designed to acquire the distance between thumb and finger at tip grasping. The deformation of the sponges caused by the pinch force at each digit’s tip is measured as distance alteration. The data were collected with the tool during the pick-and-place task on both the sound hand and body-powered prosthetic hook control simulator. The experimental results showed that the device was satisfying to measure the difference of the sound and artificial hand’s digits performing the task.

Mountaineering accidents frequently lead to lower limb injuries, particularly affecting the ankles. Ankle twisting is a common injury occurring during downhill descents. In particular, walking on cross-slopes can result in abnormal inversion or eversion of the ankle, increasing the risk of injuries. While numerous ankle assistive devices exist for walking conventional ones are often unsuitable for navigating uneven terrain. To address the challenges of cross-slope walking, we developed a dual axial ankle supporter (DAAS), featuring dual axial movement in both the sagittal and frontal planes. The Adjustable Assisting Direction Mechanism (AADM) assists the motion of the sagittal plane in DAAS. In this paper, we set a target value based on the assistance force of the mountaineering boots and confirmed that DAAS has an assistance moment that exceeds the target value. Also, Gait experiments were conducted to assess the assistive effects of the DAAS. In evaluating the impact of the AADM, goniometer signals from both ankles were measured during cross-slope walking. The findings indicate that the DAAS successfully reduced range of motion of ankle joint on the upper side of the cross-slope.

State-of-the-art teleoperation systems are installed in various specialized robots, such as disaster, underwater, and surgical robots. However, the sensitive nature of their tasks causes psychological stress in operators. To reduce this stress, bilateral controllers are used to achieve a force feedback function that can support operational skills with gravity/friction compensation and haptic perception. Therefore, we developed several types of haptic devices with magnetorheological (MR) fluids as the core technology for fine haptics. In our previous study, twin-driven MR fluid actuators were developed as torque resources for a leader-site of the teleoperation system. In this structure, two MR fluid clutches which rotate in the counter directions are utilized as high-performance torque transmitters. Additionally, three TD-MRAs are installed in a haptic interface and combined with a delta-linkage system. This delta-type haptic interface (DH-MRD) was used as a leader system of the bilateral control system. In this study, we improved and evaluated the TD-MRD and DH-MRD.

This study aims to develop a needle puncture robot to automate the pericardiocentesis procedure. In this report, we investigated whether it is possible to estimate the amount of needle deflection during puncture using LSTM, and to derive the position of the needle tip by correcting for the deflection using this information.

Care facilities are currently experiencing a serious shortage of manpower, and late-night patrols have become a major burden for caregivers. To reduce this burden, we have developed a water-sensitive urination detection device using UHF RFID without a battery and a sensor IC. In this study, we developed a method to identify and detect the amount of urination in a step-by-step manner when a urine pad is attached to a diaper in order to satisfy the demand for notification in the range of 600 ml to 1000 ml, which is the amount of urine that can be absorbed.We conducted a test to verify the accuracy of the developed method. The results showed that although the F value of the detection rate was low (68.4%), the urination detection rate was 80.0% by adjusting the number of times of notification in consideration of the detection rate.

Accidents involving power wheelchairs occur frequently among the elderly, and one of the causes is a decline in cognitive reaction time due to aging. We aim to reduce power wheelchair accidents by improving the cognitive reaction time of the elderly. For this purpose, we measured surface electromyography (sEMG), a type of biological signal that appears before movement. The subject is asked to perform an action that simulates a power wheelchair, and the surface sEMG is measured. The measured sEMGs are computed and analyzed as features for machine learning to estimate the distance and direction of the subject's hand movement. Two types of machine learning are used: SVM and Random Forest. We analyze the results using these machine learning methods and compare their accuracy.

We have been developing a human-machine interface for hemiplegic patients. We have been estimating the direction from the EMG generated during shoulder joint motion, but the identification rate was less than 50%. In this paper, we investigated the cause of the low identification rate based on the data length, the number of data, and the number of features. As a result, the number of features influenced identification. However, there were some who had an influence on the identification rate with respect to the data length and the number of data and others who did not.

This study proposes a method for estimating the forward tilt angle using an illuminance sensor, focusing on changes in the illuminance value of the seat surface caused by changes in the posture of the upper limb field during wheelchair operation. The effectiveness of the proposed method was verified by fixing the position of the wheelchair, first establishing the proposed equation by measuring the illuminance values using a board as a preliminary experiment, and then measuring the illuminance values five times per person by three healthy subjects in the same way as in the preliminary experiment and comparing the results with the proposed equation.

On trains, information on emergencies, transfers, etc. is communicated by voice, making it difficult for deaf people to recognize the information. In this research, we propose an announcement voice recognition system that focuses on the characteristics of train running sounds, and evaluate by comparing the recognition rate with voice recognition app.

Some ALS and SMA patients are concerned about the inability to communicate emergencies such as sudden changes in their health in the absence of helpers or family members. This paper presents the development of an emergency call system for a real user, which uses hand pose estimation to detect a small flexion movement of the index finger of the right hand, and then uses the results to notify the helpers or family members of an emergency. Experimental results show that real user can send notifications through the system and the time required is within an acceptable range.

The action of holding hands is said to be effective in touching used in nursing. However, the action of grasping the hand includes the action of“ being grasped ”in addition to the action of“ grasping ”. Therefore, in this study, we focused on the “ grasping ” motion and investigated to what extent the sensation of pain changes depending on the presence or absence of the motion. It is also said that there is a threshold for pain, and that the act of grasping the hand delays the onset of pain. However, since the ultimate threshold is unknown, we investigated how the“ grasping ”motion changes the pain threshold.

In periodontal examinations, precise probing force and the ability to stabilize the probe are essential skills. Practitioners must adeptly insert the probe with appropriate force, underscoring the necessity for thorough skill acquisition. To address this, we developed a probing simulator equipped with sensors that provide feedback on both probing and resting forces. This simulator allows users to independently measure and enhance their skills in probing. The incorporation of real-time feedback aims to optimize the learning process, ensuring practitioners can master the required techniques for effective periodontal assessments.

In this paper, we propose a method for multiple people tracking, identification and action recognition in a room using 2D-LDIAR at waist height. The method estimates the positions of multiple people by particle filter and, enables simultaneous person identification among 4 people and action recognition among 4 actions(standing, sitting, walking and door opening/closing). As an experiment, two people walked in a room. The RMSE of position estimation without occlusion was 0.229 m, and with occlusion was 0.307 m. These results indicate that our method have sufficient capability for human position estimation. While the accuracies of person identification and action recognition were over 90% in single person case, the accuracies in two people case showed 67% for person identification and 50% for action recognition. There is a room for improvement in multiple people case.

In this paper, we propose a system for measurement of forearm vessels deformation under loading. This system enables the simultaneous measurement of the force applied by the probe using loadcell and the vessel diameters in ultrasonographic images. In addition, the ultrasound probe can be fixed at any position on the forearm, and the load can be increased gradually by means of a fine-adjustment screw. Using a constant load spring, it is possible to measure without any additional force being applied to the forearm. Nine healthy male participants were recruited. The deformation of vessels at two locations close to the elbow were measured using this system. As a result, it was found that the vessels were twice as hard to deform and the diameter increased by approximately 1mm by tourniquet. The result demonstrated the validity of our measurement system.

We have been developing a home monitoring system in which a robot speaks to an older adult when anomaly is detected. In this paper, new stationary IoT sensor nodes were developed. The sensor nodes were powered by commercial electricity and equipped with privacy-friendly heterogeneous sensors measuring power consumption of electrical appliances, vibration, sound level, illuminance, pressure, temperature, humidity, and CO₂. We also programmed an IoT gateway to collect data every second from multiple nodes via Wi-Fi and transfer it to a database. Operational verification experiments revealed accurate measurements and uninterrupted operation for one month with data loss below 0.05% and within one minute. An anomaly detection experiment using supervised machine learnings demonstrated ability to detect various simulated anomalies embedded into accumulated data, achieving a recall rate exceeding 0.95. These results indicated the potential for the developed system to serve as a foundation for the targeted monitoring system.

In recent years, Japan has become a hyper-aged society, and accidents caused by falls and stumbles of dementia patients have been increasing. In previous research, a motion monitoring support system using depth sensors has been developed to reduce these accidents. However, among the five available depth sensors, Kinect2 and RealSense D455/L515 are no longer in production and are difficult to obtain. Therefore, we introduced a new depth sensor, TEDTOF manufactured by Tokyo Electron Device, and implemented it in the existing system. Two image filters and one image processing were used to reduce noise in the distance image obtained from the TEDTOF. In addition, a comparison experiment was conducted to discriminate the patient's condition among the five depth sensors used in this study. The results showed that TEDTOF had the best recognition accuracy and RealSense L515 had the lowest recognition accuracy.

This paper describes an introduction of a wireless auscultation sensor to improve a performance of a multipoint automatic auscultation system that has been developed. A method of introducing the sensor into the system and an analysis of the auscultation sound are also discussed. In addition, a method of attaching the sensor to a chest surface is investigated, and an attempt is made to configure a wireless auscultation sensor array.

To prevent dental caries and periodontal disease, brushing the teeth continuously over the long term is crucial so that plaque on the teeth can be removed as much as possible and gum massage with the toothbrush can be performed. To evaluate whether brushing is being performed correctly, it is necessary to assess the movement of the toothbrush and the contact force. Therefore, we have developed a toothbrushing evaluation system that measures the toothbrushing motion and its force. In this work, we implemented a function that visualizes the toothbrushes according to their contact force, and we conducted experiments using artificial plaque removal on dental models to examine the usability of this function.

Force exertion by humanoid robots requires coordinated whole-body control that takes into account reaction forces from points of contact with the environment. In this study, we generate and simulate force-actuated motions to exert force on the external environment using a wrench distribution based on VRP (Virtual Repellent Point). HRP-2, hrpsys, and OpenHRP3 will be used for the simulations. Force actions to apply arbitrary wrenches to a wall surface will be performed in both simulations using Choreonoid and OpenHRP3, and the results will be compared.

In this study, we examined the frame structure for developing a robot arm of humanoid robots from the upper arm to the forearm, based on functional anatomy. A 3D CAD model of a robotic arm incorporating human anatomical features was created, and an assessment of its range of motion was conducted. The evaluation results demonstrated that the integration of human anatomy achieved an expansion of the range of motion.

When practicing an opponent sport such as badminton, using a humanoid robot as a training partner requires the robot to be agile. To improve agility, we have used a robot operated by lightweight pneumatic actuators and equipped with suction cups to improve tipping stability. One of the challenges in quick motion of the robot is the delay in pressure response when the pneumatic actuators operate. In this study, we propose a method to apply driving force to the joints of the robot in advance, while maintaining the posture of the robot using the adsorption mechanism. As a result, the effect of the delay in pressure response was reduced and the robot achieved an increase in speed due to improved stepping out movement.

A variety of walking aids have been developed to assist disabled individuals in their mobility. However, in order to conduct clinical trials with disabled subjects, it is necessary to confirm the safety of the aids prior to actual use. This safety confirmation process can be both time-consuming and costly. As such, the development of robots as substitutes for human subjects has become a viable solution. Our research focuses on developing a small and cost-effective bipedal walking robot that moves similarly to a human. We have developed a robot foot section with hallux and fifth toe pads to mimic the plantar pressure distribution during human static standing. From the measurement results of the plantar pressure distribution in the robot foot section, it was found that as the tension of the plantar fascia increased, the load on the forefoot also increased.

This paper proposes self-collision avoidance for whole-body model predictive control (WB-MPC). Since WB-MPC requires solving a large-scale optimization problem, the gradients of the dynamics and cost functions must be computed quickly. To compute the gradient of the self-collision detection quickly, we create a collision detector using a deep neural network. The DNN-used collision detector estimates the minimum distance between all body frames based on the whole body joint angles, converting to a cost. As a result, it enables WB-MPC to plan long-term self-collision-aware motion by using its gradient calculated at high speed. The proposed method was implemented for a physics simulation using a dual-arm robot and outperformed the previous method with inverse kinematics in terms of long-term planning with self-collision avoidance.

This paper shows walking of an underactuated biped robot. This biped robot has a mechanism such that the knee joint is fixed by its own weight to the knee joint stopper when the knee joint is in hyper extension. Using this mechanism, we designed a gait that does not require knee joint torque of stance leg. The ability to walk with the proposed gait is confirmed by numerical simulations. We also designed a conventional gait such that the hips slightly lower and the knee joints are in flexion. We simulated walking with this conventional gait. As a result, the energy consumption of the proposed gait was lower than that of the conventional gait.

The Plesiosauria are a group of aquatic reptiles lived from the Late Triassic Period into the Late Cretaceous Period．They were unique because they evolved four large wing-like flippers almost identical in size and shape． Due to this unique body structure， it has been assumed that plesiosaurs may have performed a unique swimming style． However， the locomotor patterns of plesiosaurs are yet to be understood．To reconstruct their locomotion，we develop plesiosaurs-like robots and aim to understand the role of plesiosaurs trunk and verify plesiosaurs swimming stability through robot experiments． In this paper，we focus on the relationship of inter-flipper and trunk coordination in plesiosaurs locomotion because previous studies have verify flipper coordination without plesiosaurs trunk．

Paddling is a swimming styles of animals, represented by a dog paddle, and is widely used as the propulsion style for bio-inspired swimming robots. These robotics studies mainly investigate shape and movement of each paddle with a specific inter-paddle coordination, e.g., trotting gait. It is expected that flexible coordination patterns between paddles allow the robot achieve more efficient and versatile swimming. This study aims to establish a control method that generates animal-like adaptive paddling. This paper develops a swimming robot with four paddles and exams a decentralized inter-paddle coordination mechanism.

Quadrupeds exhibit surprisingly adaptive and agile locomotion under real world constraints. This ability is achieved by orchestrating many degrees of freedom in their bodies in real time. However, the control principle underlying this ability still remains elusive. In order to clarify the control principle, in this study, we focus on sprawling locomotion which is a quadruped walking gait with lateral body bending widely observed in amphibians and reptiles such as salamanders, lizards, and crocodiles. We propose a decentralized control algorithm for body–limb coordination in sprawling locomotion by fully exploiting the softness of the body. Simulation results suggest that the proposed control principle could well reproduce sprawling locomotion.

In this study, we proposed an automatic state transition algorithm through two types of interactions for a swarm that reflects sensor inputs. Each agent in a swarm is equipped with a phase oscillator whose angular velocity contains sensor input information. In the proposed algorithm, a swarm estimates the unified leader robot that has the maximum sensory input through a conventional local interaction. In addition, the candidate leader robots are always communicated to the whole swarm by a broadcast-type interaction we newly introduced. As a result, the swarm can make a distributed and simultaneous transition to the next state after a unified leader is determined. In the implementation, we verified the proposed method by introducing light as a local interaction and sound wave as a broadcast interaction to distinguish the agent with maximum input and switch state automatically.

Animals change their behavior in various environments. For example, quadruped animals exhibit various gait depending on speed. In this study, we considered the mechanism of gait generation from the viewpoint of reflexes. We constructed a control law that reproduces reflexes based on the results of previous experiments on decerebrated cats. A quadruped robot that we developed exhibited a walking gait and a bounding gait with the control law. The contribution of this study is showing that the quadruped robot changes the gait patterns only by changing the speed of hip extension. This result suggests that animals may change their gaits depending on the speed of hip extension.

We apply Asymmetric Decentralized Control Barrier Functions (Asymmetric CBF) to the collision avoidance problem in swarm robot control, and conduct collision avoidance experiments introducing symmetric CBF and Asymmetric CBF using a fully autonomous robot that carries all sensors on itself to verify the effectiveness of Asymmetric CBF in real environments.

NASA is planning robotic exploration of the dead geysers in Enceladus, a satellite of Saturn, where the life is thought to exist. This paper studies the mobility of ball-and-stick typed modular robot by changing its shape in response to uneven terrain. While the shape determination algorithm is complex due to the coordination of multiple modules in three dimensions, this paper proposes an approach by breaking it down into simpler problems. By using the reinforcement learning method, the proposed modular robot can learn the strategy to navigate itself into the unknown environment of Enceladus through repeated trials. A mobility algorithm is also under study for multiple modules, which is capable of reconfiguring in three dimensions.

This study addresses the challenge of autonomous drone flight in environments where GNSS is unavailable, using only onboard sensors like LiDAR and IMU. Focusing on obstacle avoidance without external sensors, we employ LiDAR-based SLAM for real-time 3D mapping and localization, enabling stable flight in diverse settings such as indoor spaces and densely built areas. Our method innovates by mapping environments on-the-fly and applying an extended artificial potential method for navigation, avoiding the common pitfalls of local minima. Simulation results demonstrate the drone's ability to autonomously reach target locations, underlining the potential of LiDAR SLAM for enhancing drone autonomy in GNSS-denied areas. This research lays the groundwork for future advancements in drone technology, promising wider application and efficiency in logistical operations and beyond.

In the field of robotics, the systems for robot behavior planning using video have also been proposed to deal with a variety of environments and tasks generically. The merit of this system is that it allows for generic planning unrestricted by specific policies, by outputting the generation of robot actions from the vast knowledge inherent in foundation models as video, given text input. In previous studies, the verification of action generation using video was emphasized, with object manipulation simplified for testing on simulations. This is not enough the usefulness of robot behavior planning using video for pick and place tasks. In addition, the systems from previous studies are not open, and such systems have not been sufficiently discussed. In this paper, we first focuse on utilizing Robotis’s TURTLEBOT3 Waffle Pi and Mobile Manipulator(referred to ”Waffle”) to construct robot behavior planning system using video.

Evaluation of the success or failure of actions is essential for robot systems. Evaluations should be autonomous and challenges remain like how to tackle sensor troubles. In such situations, cooperative evaluation of actions will be effective via autonomous information sharing among robots. We design an autonomous swarm robotic system that allows for information sharing between robots using natural language, based on foundation models. Then, we verify that the robots autonomously evaluate the action that a single robot cannot judge by the individual through information sharing.

In MCPP (Multi-Robot Coverage Path Planning), the entire work area is often divided into multiple smaller work areas, and the path planning for each robot in charge of each area are performed. DARP (Divide Areas based on Robot’s initial Position) is an algorithm that divides the work area equally for each robot in the path planning of multiple robots. However, in order to effectively utilize the scalability of multiple robots in actual works, it is necessary to consider an even distribution not only the area but also the work cost in the work area. This paper proposes a graph-based DARP (G-DARP) that considers the variation of work cost in the work area and partitions the area based on the work cost.

In a previous study, a peristaltic mixed transport was developed based on the transport principle of the human intestine. This system has excellent scalability because it consists of identical units, but there was room for improvement in scalability because the motion is controlled by a centralized control system. By introducing autonomous decentralized control, similar to the actual control method of human intestines, to this device, it was expected to improve not only scalability but also environmental resistance. Therefore, this study aims to improve scalability by adapting the discrepancy function, which is one of the design methods of autonomous decentralized control, to the control of this device. In this paper, we have succeeded in generating the segmented motion that is predicted to be generated by applying the discrepancy function to the device. [131 words]

Autonomous decentralized control, which is highly adaptive to the environment, is one of the organism’s means of control. Animal intestinal tracts perform different types of movements without central nervous system intervention, depending on the state of their contents. We have developed a peristaltic pump that mimics the intestinal tract. We have introduced autonomous decentralized control to this pump so that its motion changes autonomously in response to the state of its contents. In this study, we developed a mathematical model of this pump and verified the validity of the model by simulation. results of the simulation, demonstrate that the peristaltic movement and the segmental movement are switched according to the state of the contents.

Quadrupeds exhibit diverse gait patterns depending on their locomotion speed. It is believed that they select the optimal gait pattern for energy efficiency. It has been suggested that changes in whole–body movement occur during gait transitions. Elucidating this intricate mechanism inherent in gait generation is expected to contribute to the fields of biology and robotics. However, previous studies have often focused on specific animal species, leading to a lack of unified discussion regarding their mechanisms. In this study, we hypothesize that real–time adaptive coordination mechanisms occur through sensory feedback from the deformation of the soft body. As a starting point, we propose a body-limb coordination mechanism that utilizes the soft body trunk. Using a two–dimensional physical simulator developed in this study, we successfully replicated locomotion patterns such as walking, trotting, and galloping, resembling those of quadrupeds.

Insects and myriapods exhibit versatile walking patterns by coordinating their body and legs. Although some walking patterns are used only by specific species, it is interesting that they share similar gaits such as the tetrapod gait and direct wave gait. Taking this commonality into consideration, we have hypothesized the existence of common control principles across insects and myriapods. Our previous studies have shown that a single control principle can reproduce multiple walking patterns observed in myriapods. In this study, we constructed a three-dimensional simulation model of insects to verify whether the control principle proposed for myriapods can replicate the insect gaits.

Simple bipedal walking, which is a means of transportation for many humanoid robots, has problems such as poor energy efficiency and slow movement speed. As one of the methods to solve these problems, our laboratory is studying the skating motion of a biped robot. Currently, our research is aimed at enabling robots to stop faster and more consistently. However, the problem with current robots is that they frequently tip over when performing stopping movements, and there is a limit to the speed at which they can stop. In this study, I analyzed the stop motion of the robot using physics simulations and looked for ways to allow skating robots to stop more quickly and stably, even at higher speeds.