Limitation of range of motion (ROM) changes grasping style. Therefore, we proposed a method to clarify the features of grasping styles of ROM-limited hands. First, we proposed to artificially limit ROM of healthy hands by using rigid strapping tape. The palmar adduction of the thumb's carpometacarpal (CM) joint and the extension of the thumb's metacarpal (MP) joint were the target of limitation. Then, we conducted an experiment to clarify the features of grasping styles of a ROM-limited hand. For the ROM-limited hand, several grasping styles were found in trials, and there was a tendency that one grasping style was chosen finally. Grasping styles finally selected by the ROM-limited hand tend to use side region of the thumb. CM joint and MP joint of the grasping styles finally selected by the ROM-limited hand tend to be farther from ROM boundary than those temporary appeared in the trial process.

Even though a considerable number of researches have treated human gait, systems to acquire feature of human gait and analyze them only from images are yet to be completed. This paper reports the result of analysis of human gait feature extracted from RGB gait images using 3D video motion capture system. Through this analysis, some gait features made appearance by singular value decomposition. Additionally, personal identification experiment using a convolutional neural network was conducted from the video motion capture system. Although training the network contributed to correct identification to some extent, its accuracy did not exceed that of a related research.

History learning is important for children. As a child's history learning support, the authors are developing a learning system to experience early-human's life. In this system, a Kinect sensor used for a robot or the like is used. By doing this, the learner can manipulate the early-human by doing the movement he wants to make it to the early-human, not the operation with the mouse or the like. This gives a sense as if he / she is actually doing early-human's life. In this paper, we describe a system to experience hunting of early-human developed as the first step of this system.

For the construction of a body cooling system to prevent hyperthermia for spinal cord injury(SCI), the difference between persons with SCI and able-bodied persons in heat transfer between body and air is important. This paper proposes new simple linear dynamical model focused on heat transfer which can be measured from experiments. To confirm this model, experiments were conducted targeted at persons with SCI and able-bodied persons. We evaluated estimation errors of the model fitting for core temperature data were less than the limit of detection of core temperature thermometer(Core temp.CM-210, Terumo Corporation). Furthermore, persons with SCI are statistically significantly lower than able-bodied persons regarding fitting parameters related to heat transfer between body and air(p<0.05).

Muscle vibration is one of the way of gait intervention. In this research, the purpose is to evaluate the changing gait affected by vibrator using real-time gait phase detection system. We performed experiments on three normal subjects. The following muscles were tested for the effect of vibration: Rectus fermoris, Hamstring, Patellar tendon, and Tibilias anterior muscle tendon. We get gait cycle time, stride and 3 joints movement: Hip, Knee and Ankle joint. Among the four groups of muscles we studied, the largest increase in gait cycle time and stride were found during vibration of Hamstring muscles and Tibialis anterior muscle tendon .

Behavior of insect is not only dominated by central nervous systems (brain). The peripheral nerves and embodiment are also related. However, in real animals, the specific roles of the head ganglia (brain), thoracic ganglion and embodiment are still unknown. In this study, we aim to find the control law embedded in embodiment and thoracic ganglion by confirming the behavior of the head removal cricket. Experimental results show that gait pattern in the head removal cricket is different from that of intact crickets. It also shows that the gait of head removal cricket is unstable, that is, the cooperativeness of each leg is quite low.

Snakes can change their locomotion patterns in response to the environments. Inspired by this ability, we previously proposed a decentralized control scheme based on Tegotae, a concept describing how well a perceived reaction matches an expectation, and reproduced scaffold-based locomotion in unstructured environments and concertina locomotion in narrow spaces without changing any parameter. Therefore, Tegotae-based control is expected to be able to reproduce snake's various locomotion patterns. However, since we modeled the snake only in the 2D plane in the previous work, we could not reproduce the locomotion patterns utilizing anchoring to the ground such as sinus-lifting and sidewinding. In this paper, we extend the above model to add three-dimensional effects and examine whether it can reproduce the behavior utilizing anchoring to the ground by Tegotae-based control.

Most robots hardly adapt to unexpected changes in the environments and their own morphology such as failures. To tackle this problem, we have been investigating on an ophiuroid as our model, since it exhibits highly adaptive and resilient locomotion even though it is composed of primitive and decentralized nervous system. Clarifying the control principle underlying ophiuroid locomotion is expected to provide us a novel design scheme for building highly adaptive and resilient robots. Recently, we have proposed a novel hierarchical decentralized control scheme, and successfully reproduced behavior of an ophiuroid via simulation by the proposed control scheme. In this study, we build a real physical ophiuroid-like robot for validating the proposed control scheme in the real world.

Most four-legged robots move mainly by limb motions. In contrast, quadrupeds move by coordinating body and limb motions, i.e., body-limb coordination. The body plays an important role in quadruped locomotion in terms of increasing the propulsion force and extending the stride length. The aim of this study is clarifying the control mechanism for body-limb coordination underlying animal locomotion. In our previous work, we proposed an extremely simple mathematical model that well coordinates body and limb motions autonomously and successfully reproduced high speed running via two-dimensional simulations. In this study, we developed a robot to validate the proposed control scheme in the real world.

Centipedes(Scolopendridae) with a large number of legs and segments can move effectively by combining legged motion and body undulation. Clarifying the underlying control mechanism will help not only to contribute to biology but also to develop highly adaptive bio-inspired robots. However, a decentralized control mechanism that coordinates the centipede legs and the body to generate adaptive locomotion has not been elucidated. In this paper, as a first step, we proposed a decentralized control scheme that enables well-balanced coupling between the legged motion and body undulation by using a simplified model of centipede's mechanical system. We designed “bilateral” local sensory feedbacks between legs and body. In order to verify the proposed control scheme, we developed a centipede-like robot as a platform.

“Tegotae” is a Japanese concept describing how well a perceived reaction matches an expectation. Using a decentralized control system based on “Tegotae function” which quantitatively evaluates Tegotae, we have successfully developed several robots that can adapt to changes in environment. However, so far the function has been designed on a trial-and-error basic. We expect that Tegotae function can be systematically designed for various types of robot by developing an automatic generation method. In this study, we proposed an automatic generation method of Tegotae function by using a genetic algorithm and verified it via simulation using a model of earthworm locomotion.

In this research and development, we construct one-finger drive type intelligent decorative electric prosthesis that sensed finger joint motion near the defect. we considered a case where the index finger DIP joint and PIP joint were missing, devised an electric hand finger prosthetic mechanism with strong gripping power, small size, and light weight. With this mechanism, we aim to restore "pinch" and "grasp" which is the movement function of the hand. The sensing part is MP joint of the index finger or the middle finger. We control the electric prosthesis by sensing the motion in the angular direction of flexion and extension of these MP joints. As the sensing method, instead of EMG signals conventionally used, a stretchable strain sensor with high flexibility makes of rubber and urethane is used as a simpler method. In this paper, we measured the force and the movable range of this prosthesis, and compared it with motion of actual fingers and motion analysis model of fingers.

In the previous study, a grasping force observer (GFO) was proposed to estimate the grasping torque when the forceps manipulator for laparoscopic surgical robot grasped an object. However, when the tip of the forceps manipulator bent, extra load torque occurred due to the bending mechanism. Therefore, in this study, drive parts compensation function in the GFO was modified so that the load torque due to the tip bending is cancelled regardless of the bent state of the tip of the forceps. Using the improved drive parts compensation function, a compensation test for the load torque due to the tip bending was conducted under the condition that no load was added in grasping. As a result of the test, cancellation of the load torque due to the tip bending was verified.

Catheter ablation is a method to treat arrhythmias which is an abnormal contractile movement of the myocardial. While this treatment is minimally invasive, it may lead to complications such as vascular injury, cardiac puncture and cardiac tamponade. The cause of this problem is considered as the difficulty for the surgeon to intuitively operate the catheter tip. To improve accuracy of catheter ablation treatment, we developed a master-slave system that considers the structure of a human wrist for surgeons to operate a catheter remotely. A master and a slave robot made it possible to operate the catheter in 2 DoF which is the inserting and the twisting operation. Two experiments were carried out, which is to determine the insertion displacement of the slave robot and to obtain the correspondent relationship between the master and the slave robot.

In this study, a training system of laparoscopic surgery for complete beginners is proposed, so that the trainees can experience an ideal forceps operation of ligation performed by skilled surgeon. The developed training system consists of a training screen and an evaluation screen. In the training screen, the ideal forceps operation of ligation performed by skilled surgeon is displayed as a guideline, and the trainee manipulates the forceps so as to follow the guideline. Then, similarity between the forceps operation performed by trainee and the ideal forceps operation is calculated on the basis of Euclid distance of the tracking error. In the evaluation screen, the similarity is displayed visually to let the trainee know the evaluation of his/her skill. Experiment was executed to verify an effectiveness of the proposed system.

In recent years, interventional radiology (IR) which is a medical procedure has been attracting attention. Doctors can perform IR percutaneously while observing the fluoroscopic image, such as CT and MRI images, of patients. Therefore, this surgical method is less invasive. However, doctors are exposed to strong radiation in the case under CT-guidance. In order to overcome this problem, we developed remote-controlled IR assistance robot. A registration function, which calculates offsets of coordinate systems, and a three dimensional display function, which helps us to confirm the path of needle and posture of a robot,are introduced.

This paper discusses force sensing that measures a force applied to the tip of a wire-driven multi-degree-of-freedom forceps with tension measurement of the wires. The wire is beat and vibrated to generate sound. The tension of the wire can be calculated from the sound frequency by using the vibration theory of strings. Measurement of multiple axis forces is also possible with only one microphone. We develop a forceps that integrates a beating mechanism and experimentally confirm the change of the sound frequency as the change of the wire tension.

Inertial measurement unit (IMU) was used to measure walking. There remain the problem that the result using IMU was affected by accumulate error. To solve this problem, digital human model was used as constraints to compute walking posture. Center of mass (CoM) during walking was estimated by using the digital human model and the IMU. The IMU were attached on lower limb of subject. The measurments of the individualized digital human model was utilized to compute position of each joints. The posture of each lower limbs were obtained from IMU. The estimated CoM were comparing that measured by the optical motion capture.The estimated CoM trajectory showed similar pattern of the trajectory measured by optical motion capture. The largest error was confirmed in anterior posterior direction. This result suggests that accumulate error have less effect by using digital human model.

In this paper, we propose the inverse kinematics using the high-order information moments and build video motion capture system with multiple cameras. The high-order information moments are what combines the high-order moments from Part Confidence Maps showing the existing probability of joints in an image with ones from a human silhouette image. In this system, there're two steps. First one is rough estimation and the next one is precise estimation. Rough estimation is estimating 3d position of major human joints. Precise estimation is estimating 3d position of other parts including hands and feet. For rough estimation, we achieved 6.76cm about the average joint error over 740 jumping motion. For precise estimation, by dividing a whole body into some parts, we built the technical basis about video motion capture considering a whole body including hands and feet.

In this research, we established a training method of juggling performance in order to improve juggling skill. For realizing that, we developed real-time auditory feedback system while juggling training. In this system, it calculated the average of individual throwing intervals in real time and feeds that back as auditory stimuli. For evaluating that, we compared with and without using system in 10 minutes training experiment. As a result, it was found that coefficient of variation decreases with feedback in both groups. The reason for this, when using the auditory feedback system, we found that it is possible to follow the sound even if it is not consciously matched to the interval of the sound. Therefore, it was concluded that the training system in this research was effective.

In natural and social systems, decentralized control plays an important role for generating ordered structure. However, the design principle of decentralized control that is applicable to various systems has not been established. One reason for this is that it is not understood well how a nontrivial macroscopic behavior emerges through interactions among microscopic behaviors of individuals. In this study, we introduce a Japanese concept sontaku, which means “consideration for others”, to design a decentralized control scheme that enables to generate well-coordinated global function. In this paper, we investigated the validity of sontaku-based decentralized control by applying it to legged locomotion.

This paper focuses on a decentralized multi-robot patrolling system. In this system, the patrolling mission is to detect visitors as many as possible. For this mission, patrolling task assignment is a challenge. Therefore, we propose a territorial approach without using communication among the robots. In this approach, the target environment is divided into two territories when a robot meets another one. This division process is done by expanding the occupancy area between the robots. After that, shapes of the divided territories are modified on the basis of the distance and workload. Through the multi-robot patrolling simulations, we show the effectiveness of the proposed territorial approach.

Insects exhibit various locomotor patterns in response to the environment encountered. Such locomotor patterns are generated via coordination between leg movements, i.e., interlimb coordination mechanism. If we clarify the interlimb coordination mechanism, it will definitely contribute to the realization of legged robots with high adaptability to working environments. In this study, we aim to understand the interlimb coordination mechanism of hexapod walking by drawing inspiration from that of myriapod (such as centipedes). From the viewpoint of evolutionary process, insects are descendants of centipedes, and thus there likely exists a common principle underlying both species. We investigated the validity of the control scheme by using a hexapod robot, and successfully reproduced in part hexapod locomotion patterns.

Several animals such as snails and earthworms exhibit crawling locomotion by propagating the bodily wave of expansion and contraction to adapt to various environments. In this locomotion, not only expansion and contraction of the body but also friction with the environment is considered to be an important factor. However, existing studies has not sufficiently integrate peristaltic movement with friction control. Therefore, in this research, we propose a decentralized control of crawling locomotion that well-integrates peristaltic movement with adaptive friction control, and simulation results represent that a crawling locomotion with high environmental adaptability was realized.

Multi-agent reinforcement learning (MARL) is a framework to make multiple agents (e.g., robots) in the same environment learn their policies simultaneously using reinforcement learning. In the conventional MARL, although decentralization is essential for feasible learning, rewards for the agents have been given from a centralized system (named as top-down MARL). To achieve the completely distributed autonomous systems, we tackle a new paradigm named bottom-up MARL, where the agents get respective rewards. The bottom-up MARL requires to share the respective rewards for creating orderly group behaviors, and therefore, methods to do so were investigated through simulations. We found that the orderly group behaviors could be created by considering the relationship between the agents.

We propose a mesh robot based on Computer Graphics. This robot aims to hold and deform three-dimensional curved surface. Furthermore, it deforms the curved surface even by external force, and changes the ease of deformation for each part. Since Circle Packing is used for coordinate transformation from a plane to a curved surface, the module shape is assumed to be a circle. Based on the method proposed in the field of mathematics, simulation of Circle Packing Mesh's initial preparation was carried out. We established the equation of motion of the mesh robot as a set of rigid disks. We solved the equation using BiCGSTAB method.

Heart rate is one of the biological rhythms and has fluctuating properties. By analyzing the fluctuation, we can obtain useful information on the health state. However, because heart rates are usually measured with electrodes, the measurement of instantaneous heart rate, which has equivalent information to heart rate variability, is also difficult to be conducted on a daily basis. To solve this problem, we are developing systems to detect instantaneous heart rate from output of a soft rubber-based tactile sensor. We may obtain instantaneous heart rate from the sensor output by using a signal processing technique called complex demodulation. To verify the accuracy of this method, we measured ECG and pressure time series from the tactile sensor simultaneously, and compared the detected instantaneous frequencies using cross-correlations. The results were quite promising. This shows the possibility that we may be able to develop a nonrestraint health monitoring system using a tactile sensor.

Photoplethysmographic (PPG) signals contain body motion artifacts. In order to use PPG signals for daily basis pulse rate measuring, we must cancel body motion artifacts from PPG signals. In our study, we made a prototype board for PPG measuring. This board also contains 3-axial accelerometer for body motion measuring. This board was affixed to the upper arm to measure signals. After measurement, pulse rate was detected by using offline signal processing to reduce body motion artifacts, where signals from 3-axial accelerometer were used. After processing, we succeed in obtaining more accurate pulse rates.

Echography is inexpensive and indispensable equipment in every medical field but examination with Echography greatly need the expertise and experience of an operator. So, person who is not skill in echography not easy inspect. We have constructed an echography support system for reducing the burden of doctor and patient so far. We estimated the mental rotation matrix by focusing on the echogram change and probe movement. We focused on the acoustic window can be taken clear images avoiding bones and organs. Also, experts have focused on placing probes on the acoustic window and have been working on building probe contact position and orientation measurement systems. We set a region of interest called ROI in the obtained echogram, and analyzed the average luminance value in the region. In this paper, we report estimation and measurement of the probe movable range which acquires only the acoustic window by combining the analysis result of the average luminance value with probe contact position information. We propose system for examination acoustic window by estimating range of motion of probe for examination with desired echography.

Among various medical problems facing us in today's Japan are rapidly aging society, localization of medical doctors, concentrated visits of patients to large hospitals, etc. We have developed an ultrasound diagnosis and treatment support system for reducing the burden of doctor and patient so far. We succeeded in the tracing of the organ which moved a dislocation image by body movement or breathing. However, the ordinary ultrasound probes, for tracing the three-dimensional movement of the organ has been difficult. Therefore, compensation for body movement has been needed. In this paper, we performed experiments using gallbladder by echogram processing software and measured the distance the center of the organ for proposal of visual feedforward control system.

Quadruped animals exhibit variety gait patterns, and they have an anterior-posterior asymmetric structure in their their body. This study aims to clarify what kind of influence of the asymmetry between their front and hind legs on their gaits of quadruped locomotion. We built a quadruped walking robot simulator using Open Dynamics Engine (ODE). Each leg of the robot was controlled by independent phase oscillators and implemented TEGOTAE-based control, which could generate a variety of gaits by using the interaction between its body and the environment. Results of walking simulations conducted changing its leg length show that the transitions of leg phase is different based on the asymmetry.

Currently, robots are used for pipeline inspection since it is more convenient than manual inspection. In this paper, we proposed a concept of wireless relay communication to assist robot to extend the inspection range. Specifically, we developed a tracked robot-chain system. Leakage information of pipes are transmitted through multiple robots. To ensure the stability of relay communication between adjacent robots, we adopt RSSI-based evaluation method. We conducted some experiments to evaluate relay transmission efficiency, RSSI-based cooperative movement, and comprehensive performance. The results revealed that our proposed robot chain system could realize relatively accurate data relay transmission and basically RSSI-based coordinated movement.

In this paper, we develop a three-legged robot without a brain. This tripedal walking robot generates gait patterns adaptively and autonomously with no sensor, or controller. The key of this adaptability is to utilize a “weakness” of low-torque motors in each leg. Each motor drives leg joint and adjusts its own phase in response to a reaction force from the ground. As a result of the dynamics of body–motor–environment interaction, the weak motors sync each other.

In our previous work, we proposed a walking pattern generation method of a four-legged robot using four-coupled van der Pol (VDP) equations. We use VDP equation as a nonlinear oscillator to generate a walking pattern of the robot. When several equations are coupled mutually, they influence each other and their periodic solutions synchronize. Our walking pattern generation method can control forward and turning motions of a four-legged robot by changing a phase difference pattern between VDP equations. In this study, we describe the numerical simulation result that analyzed tracking characteristic to a target pattern of a phase difference pattern in each case of all or serial coupled VDP equations.

We are developing a modular robot for multi legged walking robot with two legs per module. Modular robots that we have produced so far were connected by joints between modules. It was difficult to generate a gait when connecting a large number of modules. Therefore, we newly developed a modular robot that connects modules by magnets. In this paper, we describe the hardware of the module robot which was created to generate the gait by independent CPG.

In nature, there are many insects with obvious different length of their legs. However, there was not much discussion about the influence of these leg length differences on walking. In this research, in order to elucidate the influence of the difference in leg length on walking, we focus on locusts, whose hind legs are considerably longer than other legs. Experimental observation shows that an inherent locust walking are quite different from other insects. Based on the result, we model locusts and simulate its locomotion using ODE. We realize walking locusts using the Tegotae-Based Approach and analyze the achieved walking.

For service robots operating in complex indoor environments, intelligent sensor based navigation is key for autonomous operation. In this paper, we present semantic navigation for service robots operating in indoor environments using range and vision sensors. We propose a novel mapping and navigation algorithm using topological and metric mapping. For semantic navigation a dictionary of visual objects is stored that allows the system to distinguish between different sections of the map easily. This enables to send easy human like command based navigation goals for the service robot. The proposed method is tested in simulation environment and the results are discussed. One advantage of the present system is that it does not require a large amount of memory for storing the map and based on the application the system can easily be switched between either the topological or metric map.

We propose a resetting method with GNSS for a LIDAR based Monte Carlo Localization (MCL). This method combines the GNSS resetting method and the expansion resetting. LIDAR based MCLs are widely used for mobile robots. However, MCL has a problem referred to as the kidnapped robot problem which means dissociation of the estimated pose and the true pose by an unexpected problem. Though our previous GNSS resetting method solves this problem, this method has inefficiency toward short-range dissociation. We newly propose to add the expansion resetting method to the previous method so as to make the method enhance the efficiency. The method is evaluated with the log data obtained by an actual outdoor mobile robot.

This research aims to develop a mapping for a wheeled inverted pendulum robot. Using a combination of 3D laser range finder data and odometry data an occupancy grid map is generated. It was determined that lag between the data sources causes distortion. In this paper, a method of improving the generated map based on estimating and compensating the lag is presented. In addition, a method of increasing map information based on a new grid classification is proposed and its effectiveness is verified.

We describe recognition of deviating from the route method by monocular camera image. Recently major self-position estimation methods of autonomous mobile system(AMS) use accurate map and observation information (distance, coordinates, etc.). However, these has some problems, high generating map cost, recognition miss matching and deviating from the route etc. We propose the self-position estimation method like human. This method uses monocular camera image. AMS detection distinctive features in surrounding environment from image and estimate self-position. Moreover, when it is different from the assumed route, it returns to the correct spot and route using parameter of correctness of self-position. The result of a simulation shows this method is possible to return correct route.

In order to realize semiautomatic driving system of an electric wheelchair, a dynamic window approach is used as a path control. In addition, our system input to each right and left wheel controller for desired velocity is set. We conducted a simulation in an environment where obstacles exist using this method. As a result of the experiment, it is possible that the electric wheelchair reach a target point while avoiding the obstacle.

In this paper, we propose a novel method ”laserVAE” for learning feature descriptors of scan data from a 2D LIDAR, which are suitable for self-localization of a mobile robot in an environment. Our laserVAE is an enhanced version of variational autoencoder, which is tuned up for managing step-edges in scan data. Through experiments in a real environment, we demonstrate the effectiveness of the proposed method.

We will proposed a Shape Locking Mechanism following the deformation of Flexible Rod. This mechanism is inspired by the deformation of the coil spring. Torsional deformation of the coil spring is reproduced by rotation of the joint part of the locking mechanism. This mechanism corresponds to the deformation of the flexible beam by reproducing the deformation of the coil spring. By stopping this rotation, it is possible to lock the shape of the flexible rod and switch the stiffness.

Various types of soft pneumatic actuator have been developed. However, most of these designs are limited because they are made by molding. A Stratasys 3D printer can print soft and hard material accurately at the desired positions and we can make complex structures that are difficult to make by molding. This advantage can distribute viscoelastic elements to suitable positions. Basing on this idea, we design 3-DOF actuator in this paper. We demonstrate validity of the design methodology through some experimental results.

We develop variable rigidity devices capable of handling curved surfaces, switching at high speed, and driving with low energy consumption. Also, we will prepare countless very small locking mechanisms with very simple structure, and think that one minute locking mechanism will fix it to the shape of a part of the form frame very much. In the future, it is necessary to perform simulation and optimization in order to determine various parameters, such as the method of arranging band-shaped elastic bodies, the initial state of buckling, the selection of the actuator and the speed of jumping buckling phenomenon.

This study develops a new flexible mobile robot using hydrodynamic framework. In our previous study, we developed ciliary actuators capable of generating driving force for the mobile robot. The ciliary actuator operates with one air chamber of bag-like structure. This paper describes a new omnidirectional mobile robot with the ciliary actuators arranged in multiple directions. It is expected that the robot inspects a narrow space that human beings cannot easily enter. To increase mobility performance, this study proposes a lifting mechanism that selects only necessary actuators for a specified movement. This paper demonstrates that the robot successfully moves in all directions and rotates on the spot.

An octopus has multiple degrees of freedom and elastic legs that can skillfully perform various actions. In this study, we propose an octopus type robot with multiple limb, multiple degrees of freedom and the adsorption function. We made a prototype robot with multiple degrees of freedom legs. We also carried out experiments about rod climbing, ladder climbing, brachiation, and caterpillar type movement using the robot. Because the link length of the multiple degrees of freedom leg is shorter than the link length of a fewer degrees of freedom leg, the robot has small moment arm. These experiments show examples of various behaviors by the robot with small moment arm.

In this paper, we develop mobile body which flexible deformable formed using 3D printer, and a purpose at realization of omni-directional movement operation. The bending module formed using 3D printing expands and bends by pressurization of the air chamber. Traveling waves are generated by a combination of bending motion of modules by pressurization and move operation is performed. In addition, by combining traveling waves that intersect vertically, moving motion in all directions is realized. Based on the experimental results, we confirmed moving motion in all directions by using traveling wave. An omni-directional movement operation was realized in the development of a planar flexible deformable moving body combining a plurality of bending modules.

Mutual entrainment phenomenon is also called synchronization phenomenon, if various synchronization patterns of mutual entrainment can be controlled freely, it is considered to be applicable to vibration control of multiple degrees of freedom system. In our previous work, we showed that the energy of a swing system can be controlled by periodic input control method, and that forced entrainment occurs in the system. In this paper, we describe the result that analyzed by simulation and experiment whether mutual entrainment occurs when two swing systems are coupled mutually and show that periodic motions of the coupled swing system are controllable by our synchronization pattern guide control method.

For autonomous mobile robots, self-localization is an important problem. In order to improve its robustness, we have proposed a sensor fusion system, where existent probability distributions with fault sensor information are detected by calculation of similarity between probability distributions. The usefulness of the proposed method was shown throught experimental results of position estimation under real outdoor environments. In these experiments, the robustness of position estimation was inproved by elniating information of the sensor which did not work in the environment. In this paper, we study on the robustness of the position estimation in such a environment as one of sensors does not work in the middle of the robot movement.