Information about environmental stiffness is useful for robotic tasks involving interactions with unstructured and unknown environments. However, online estimation remains a challenge. Owing to the nature of its calculation algorithm, a large amount of noise may be generated, depending on the response value of the force and position. In this study, we propose a variable gain filter that predicts the degree of such noise using a probabilistic approach and reflects only reliable data in the estimation. We show experimentally that the proposed method improves the accuracy of the stiffness estimation without degrading the estimation time constant.

Robotic motion generation methods using machine learning have been studied. Bilateral control-based imitation learning can imitate human motions using force information. Using this method, variable speed motion generation that con-siders physical phenomena such as the inertia and friction can be achieved. However, the previous study focused on a simple reciprocating motion. To learn the complex relationship between the force and speed more accurately, it is necessary to learn multiple actions using many joints. In this paper, we propose a variable speed motion generation method for multiple motions. We considered four types of neural network models for the motion generation and determined the best model for multiple motions at variable speeds. Subsequently, we used the best model to evaluate the reproducibility of the task completion time for the input completion time command. The results revealed that the proposed method could change the task completion time according to the specified completion time command in multiple motions.

This study proposes a technique for determining completion of robotic tasks considering a characteristics that some features on force responses are common among various objects. In particular, this paper focuses on the click response because its pattern is common to many objects, although the magnitude of the response differs depending on the object. A discriminator for detecting the click response based on the force sensor values, which combines mel-frequency cepstral coefficient and time-delay neural network is introduced. Based on this discriminator, how to improve the generalization performance to untrained objects is investigated to detect click responses in general for various objects. The experimental results show that the performance can be improved by including several kinds of objects with close time constants and different amplitudes of click responses in the training data.

The following topics are dealt with: mobile robots; learning (artificial intelligence); robot vision; path planning; motion control; medical robotics; optimisation; object detection; position control; collision avoidance.

In this paper, we discuss the use of pulse width modulation (PWM) with high-carrier frequency as an input for functional electrical stimulation (FES). Previous studies have shown that the relationship between muscle inflow current and joint exerted force can be approximated by a first-order lag system that includes the dead time. This characteristic can be improved by using an input based on high-frequency PWM. In PWM, the component of the carrier frequency can be made smaller than pulse amplitude modulation (PAM), and in addition, the carrier frequency can be set to a high value. This means that it is possible to move only the frequency component of the desired operating command without any effect of the carrier frequency. First, PWM and PAM were experimentally compared. Second, the performances achieved by using high-carrier frequency and low carrier frequency were compared. Finally, the control performance achieved using a sinusoidal motion command was verified by changing the carrier frequency. The gain attenuation caused by the non-linearity of the human body was clarified for high-carrier frequencies. Therefore, the control performance was worse than the originally desired result. However, the advantages of using high-carrier frequency to reduce fatigue attenuation and to suppress the effect of carrier frequency were demonstrated.

Currently, in the field of FA(Factory Automation), automation has been accomplished for only tasks with high reproducibility. Meanwhile, there remain a range of areas where tasks cannot reliably be performed due to their complexity or larger environmental variation. In these cases, the reliability of the task can be improved by recognizing the task failure through success/failure identifications and executing the task again when it fails. However, the success/failure identification using conventional machine learning methods has not been discussed for determining the success or failure for unlearned objects. Thus, this paper examined assembly tasks and demonstrated that the success or failure for an unlearned object can be identified by taking advantage of generalized nature of the neural network using force information. The results of making success/failure identifications using information on force, image, and position were compared and the advantage of force information in tasks was demonstrated.

This paper proposes a compact High Dynamic Range (HDR) six-axis Force/Torque(F/T) sensor. It is important for robots to acquire environment information based on accurate force sensing to perform operations in various environment. However, it is a problem that no compact size F/T sensors with high dynamic range exist. Therefore, a compact HDR six-axis F/T sensor with a new cross-arch structure was developed. The F/T sensor showed high linearity with respect to force component and it was possible to remove the inter axis interference component by calibration. It was also shown that this F/T sensor detect 0.2N to 500N force.

Various adaptive abilities are required for robots interacting with humans in daily life. It is difficult to design adaptive algorithms manually; however, by using end-to-end machine learning, labor can be saved during the design process. In our previous research, a task requiring force adjustment was achieved through imitation learning that considered position and force information using a four-channel bilateral control. Unfortunately, tasks that include long-term (slow) motion are still challenging. Furthermore, during system identification, there is a method known as the multi-decimation (MD) identification method. It separates lower and higher frequencies, and then identifies the parameters characterized at each frequency. Therefore, we proposed utilizing machine learning to take advantage of the MD method to infer short-term and long-term (high and low frequency, respectively) motion. In this paper, long-term motion tasks such as writing a letter using a pen fixed on a robot are discussed. We found differences in suitable sampling periods between position and force information. The validity of the proposed method was then experimentally verified, showing the importance of long-term inference with adequate sampling periods.

This paper describes the design of a high dynamic range (HDR) six-axis force/torque sensor. The sensor is composed of a high-rigidity flexure element detecting large force and a low-rigidity flexure element detecting small force. The overload on the low-rigidity flexure element is prevented by an overload protection mechanism. An HDR measurement is achieved by combining the outputs of the two flexure elements. A loading test for the designed sensor is performed, and the results indicate that the six-axis sensor measures force with a dynamic range from 0.01N to 1000 N.

Bilateral control, a remote-control technique, is used to work at a distance. However, many existing bilateral control systems have two common problems: 1) it is difficult to create a system like a human hand, that has multiple degrees of freedom and 2) if the mechanism becomes too complicated, operators feel restrained and experience discomfort. Because, for these reasons, the bilateral control of fingers has not been accomplished to date, we aimed to overcome this by applying functional electrical stimulation (FES). In our experiments, through an adhesive electrode pad, electrical stimulation was delivered to the muscles that flex and expand the metacarpophalangeal joints of the thumb and middle finger. Position-symmetrical bilateral control was implemented so that the deviation of the master's and slave's positions relative to each other was zero degrees. A sliding mode controller was used as a position controller. We found it possible to control multiple degrees of freedom; however, we found areas where the number of tracking errors was large. We speculated that the middle finger did not bend, because the arm rotates as the thumb was abduction, therefore the position of the motor point of the middle finger deviates from the position of the pad.

This paper proposes high backdrivability control which can be applied to R/C servo motors. This control method is constituted by shaft torsion control and admittance control. In shaft torsion control, the motion of the robot is derived from shaft torsion. Shaft torsion is estimated by using Shaft Torsion Observer (STOB). On the other hand, in admittance control, the motion of the robot is derived from reaction force. The reaction force is estimated by using Reaction Force Observer (RFOB). These control methods are selectively used depending on the movement of the motor. By using the proposed method, back-drivability of the R/C servo motor was improved in comparison with conventional admittance control.

In this study, we developed a method to estimate the relationship between stimulation current and volatility during isometric contraction. In functional electrical stimulation (FES), joints are driven by applying voltage to muscles. This technology has been used for a long time in the field of rehabilitation, and recently application oriented research has been reported. However, estimation of the relationship between stimulus value and exercise capacity has not been discussed to a great extent. Therefore, in this study, a human muscle model was estimated using the transfer function estimation method with fast Fourier transform. It was found that the relationship between stimulation current and force exerted could be expressed by a first-order lag system. In verification of the force estimate, the ability of the proposed model to estimate the exerted force under steady state response was found to be good.

This paper proposes a method to improve the accuracy of the force/torque (F/T) sensor based on machine learning considering time series data. There are several problems with F/T sensors, one of which is hysteresis. Hysteresis is a factor of error dependent on force history. There have been few researches focusing on hysteresis in an F/T sensor. We solved this problem by considering time series data. Time series data was put into machine learning such as linear regression and Support Vector Regression (SVR). We evaluated this method with an existing high dynamic range F/T sensor. We confirmed that the error decreased in both high and low force ranges. Since there is nonlinearity in hysteresis, we predicted that SVR will be more accurate than linear regression. Linear regression considering time series was better than SVR when loading training data at random intervals and loading test data at constant intervals.

This study proposes an imitation learning method based on force and position information. Force information is required for precise object manipulation but is difficult to obtain because the acting and reaction forces cannot be separated. To separate the forces, we proposed to introduce bilateral control, in which the acting and reaction forces are divided using two robots. In the proposed method, two models of neural networks learn a task; to draw a line along a ruler. We verify the possibility that force information is essential to imitate the human skill of object manipulation.

This paper proposes a trajectory planning method with neural networks that learn model predictive control for dynamic manipulation. The novelty of this method is that target positions and model parameters are added to the input of the neural network. The proposed method solves the dynamic trajectory planning issues with low calculation cost. This paper shows the effectiveness of the proposed method through demonstrations of a robot turning a pancake over under various parameters. In addition, we verify a performance change with respect to the number of training data.

This paper describes the reduction of chattering in functional electrical stimulation (FES) with Smith compensator. We proposed a bilateral controller using FES. In the bilateral controller, in addition to normal remote control, the reaction forces of the distant location are transmitted back to the operator. Therefore, the operator feels like being in the distant location. In controlling of the human body using FES, a dead time occurs. As a result, high-frequency vibration (called chattering) occurs. In this paper, we propose a control method suppressing chattering in FES. In the experiment, the performance of the proposed method was verified by target value control and bilateral control. In the target value control, chattering was suppressed by the proposed method. However, in the bilateral control, chattering was not be suppressed by the interference of the time delay of the two subjects.

This paper proposes intrinsic contact sensing using a particle filter. Intrinsic contact sensing is a method, which detects the contact position without any sensors on the outer shell. Force sensors are attached on the supporting point of outer shell instead. This method has been expanded to estimate contact information even without shape information. However, previous methods assume that contact points are stationary. Therefore, in this paper, we consider contact points with movements. When contact points move, it may be nonlinear. The particle filter is used to correspond to nonlinear motion. Experimental results show that the proposed method can follow the sharp movement compared with the conventional method.

In dynamic manipulation, robots can manipulate objects without grasping by utilizing inertia effect. However, the trajectory planning for dynamic manipulation is a difficult issue due to dynamic constraint. Trajectory deformation considering dynamic constraint after original trajectories are generated is necessary for the issue. To realize such deformation methods, we introduce on sequence-to-sequence (seq2seq) models, which can convert a time series to another time series. This paper proposes a trajectory deformation method with seq2seq models deforming trajectories to satisfy dynamic constraint. Users can obtain trajectories for dynamic manipulation by designing outlines of motion and inputting them to the proposed seq2seq model. In addition, this paper proposes a learning curriculum that does not need labeled dataset. Only mathematical representation of constraint and unlabeled trajectories are necessary. We implement a seq2seq model by the proposed method to a robot turning over pancakes and confirm the validity by a simulation and an experiment.

This paper introduces a method of posture control for jumping robot on the frontal plane. Human beings can perform the dynamic motion (e.g. running, jumping) by taking an advantage of muscle elasticity. Biped robots with this feature, therefore, have been studied, recently. Especially among them, many robots implemented pneumatic artificial muscles to adopt this feature. These robots have achieved the jumping motion or running motion. Although, a stable dynamic motion has not been achieved yet. Therefore, a control for the stable motion is desired. In the jumping motion, many researches have verified the stabilization on the sagittal plane. Nevertheless, these robots have fallen down after some jumps. One reason is that they cannot deal with the disturbance on the frontal plane. In this paper, a frontal posture control of a jumping biped robot is proposed. This control makes difference between the stiffness of both legs depending on an inclination and an angular velocity of the robot. To verify the effectiveness of the proposed method, we propose a model using this method. Then, this model is examined by a jumping simulation with various disturbances. The results show that this model with the proposed method improves the robustness against the large disturbance.

Electro-hydrostatic actuators (EHAs) are said to have potential as actuators for robotic systems because they have high power-to-weight ratio, compact, and energy saving systems compared to general hydraulic systems. However, modeling approaches and analyses on dynamic characteristics have not been quite concrete yet. In order to expand EHAs' potential to be used for robots, further studies of their characteristics are necessary. The previous research has shown two-inertia resonance would occur in EHAs and EHAs can be modeled as two-inertia systems. Nevertheless, how the rigidity of hydraulic hoses relates to resonance frequency and efficiency of EHAs has not been discussed. The rigidity of hydraulic hoses has been said to have an impact on resonance characteristics of general hydraulic systems. However, dynamic characteristics of EHAs and analyses on the rigidity of hydraulic hoses have not been clearly examined. Thus, this study investigates the effect of the rigidity of hydraulic hoses by applying different type of hoses in an EHA.

Some stroke patients are affected by hemiplegia, which is the complete paralysis of half of the body, and need to train in upper limb motions with high frequency to improve their quality of life. Therefore, the authors developed a portable robot with a main processor, actuators, motor drivers, a force sensor, and a touch display to assist such high-frequency training at home. To recover the motor function of the upper limbs by voluntary movement, this robot instructs the joint motion based on the force input. Therefore, this study proposes a force control algorithm for the omni-wheel mechanism and a method that estimates the odometry of the omni-wheel mechanism based on the circuit equation of the DC motor without rotary encoders. Moreover, a method that estimates integrated electromyograph patterns based on the force and movement of the robot using a multiple linear regression model is proposed for the evaluation of co-operative movement. The effectiveness of these proposed methods were confirmed by evaluation experiments.

When robots interact with the environment, it is necessary for them to detect and control contact forces. In many cases, force/torque sensors or tactile sensors on robots do not contact the environment directly, but instead contact through end-effectors or grasped objects. Thus, robots have to estimate actual contact forces through insensitive objects. However, estimating multiple contact forces through insensitive objects is an ill-posed problem. This paper proposes a method of estimating contact forces on three points separately with a six-axis force/torque sensor on an end-effector. We discovered that the components of contact forces normal to the plane defined by the three contact points can be found uniquely by applying static equilibrium of force and moment. The tangential components of contact forces can also be determined by the use of an additional algorithm. The validity of the proposed method is confirmed by experimental results.

Servo valves are unnecessary in electro-hydrostatic actuators (EHAs) because they are directly controlled by servo-motors. Owing to the above characteristics, the control frequency and power efficiency improve, and the size of a hydraulic system reduces. However, oil leakage and static friction tend to be significant in EHAs. In addition, the nonlinear characteristics of oil leakage make compensation difficult. In this study, we propose an EHA model that reflects oil leakage and static friction. To eliminate the necessity for complicated leak identification, oil leakage is estimated using two encoders. Then, the effect of static friction is suppressed using feedback modulators (FMs). Finally, disturbance observers (DOBs) are integrated into the model to eliminate disturbances in the EHAs and the validity of the proposed method is verified experimentally.

This paper deals with a tactile interface using force sensors. Touch panel, which uses contact position information for input, is spreading as a good candidate of intuitive man-machine interfaces. In addition to contact position information, using contact force information realizes intuitive and versatile input. “Haptic Desk” is a desk-type tactile interface that force sensors are attached to the legs of the desk. This technology can estimate contact force and contact position on the desk. Moreover, this technology can convert not only a desk but also other products, such as consumer electronics and furniture, into tactile interface. However, if the products are vibrated, the errors of the estimated external force and contact point occur. Because the inertia force of the product is not considered, the errors occur. For reduction of the errors, this paper expands the method of Haptic Desk by using an acceleration sensor. The usefulness of the proposed method is shown by experiments.

This study investigates bilateral control between electric and electro-hydrostatic actuators (EHAs) using a feedback modulator (FM). In bilateral control, we apply a local robot as a master driven by an electric actuator and a remote robot as a slave driven by an EHA. EHAs exhibit a high power-to-weight ratio with low energy loss. However, there would be difficulties when it comes to control designs because of non-linear characteristics such as friction. In order to exhibit high control performance of bilateral control, the effect of static friction which occurs in the hydraulic actuator of the slave has to be eliminated. An FM is applied to reduce the effect of static friction of the hydraulic systems and to enhance responses of actuators. We apply the FM to an oblique-coordinate-bilateral-control system and investigate the control performance. The experimental results indicate that the FM reduced the effect of static friction and the position control performance was improved. Besides, the conventional method is not able to transmit the reaction force because of the large static friction in the low-speed domain. However, the proposed method stabilizes the contact motion by applying the FM.

This paper introduces a method of force control for a jumping biped robot to correctly control the ground reaction force during continuous jumping. Assuming dynamic motion, such as jumping and running, the attitude of the robot depends on the dynamics in response to the ground reaction force. Therefore, controlling the ground reaction force is necessary for stabilizing the robot's motion. However, control of the ground reaction force involves a fundamental problem: feedback control does not work properly against the impact force owing to limitations in the control bandwidth. This study introduces a method for stiffness ellipse control that utilizes three antagonist pairs of six pneumatic artificial muscles. When the tip of the robot makes contact with the ground, an external force is induced in the direction of lower stiffness. By utilizing this property of the stiffness ellipse, it is possible to control the ground reaction force by feedforward control, which is not dependent on the control bandwidth. Based on this idea, impact force control at landing and jumping force control at takeoff were proposed to correctly control the ground reaction force during the continuous jumping of the robot. The results of several experiments conducted convince us that the relationship between the ground reaction force and the stiffness ellipse is almost linear, and that the ground reaction force can be controlled with high reproducibility by adjusting the stiffness ellipse.

Computer vision plays an important role in various applications. However, the accuracy of computer vision is often degraded by specular reflection. To solve this problem, a system utilizing a high-speed camera and a strobe for removing specular reflection has been shown to be effective. This method utilizes a principle of estimation using images with luminances varied by the flickering of a strobe. However, the method generates noise due to the difference in position of specular reflection among images in video imaging. In this paper, the authors propose a method combining optical flow with the above-mentioned method. The proposed method enables the removal of specular reflection for video imaging without generating noise. Experimental results and comparison with other methods of specular reflection removal show the effectiveness of the proposed method.

This paper describes bilateral control using functional electrical stimulation (FES). Bilateral control is a teleoperation system using force feedback. If we can use it in everyday life, it could allow remote diagnosis by doctors and communication of tactile information. However, in conventional studies of bilateral control, operators must wear robot arms. Using them in everyday life is not desirable, because they are restrained and make human feel uncomfortable. Therefor, we propose the use of FES in the transmission of force information. An experiment on bilateral control of elbow joints with five healthy male participants was conducted. Biceps brachii muscles and triceps brachii muscles were stimulated using adhesive electrical pads. Position symmetrical bilateral control was implemented so that the deviation of the position of each other is 0. The angles of the elbows were measured by encoders, and a proportional-integral-derivative (PID) controller was implemented as a position controller. The experimental results show that the proposed method showed good performance in the practical viewpoint. However, the elbows of some subjects could not be successfully controlled because interference current occurred between their muscles.

This paper focuses on extending a previously reported technique to convert a desk into a tactile interface. The original technique, based on the so-called "Haptic Desk" enables the detection of an applied force and its contact point by force sensors at the desk's supporting points. Consumer electronics and furniture can be converted into tactile interfaces with the use of this technique. However, this method has the restriction that an object should be composed of planes. The authors introduce a method of modeling the object as a set of points and estimating the contact point based on the distance between a point and the force representative line. Since there is a possibility that an incorrect contact point is estimated, the standard deviation of the distance is used to prevent incorrect estimations. The results of experiments show that the error is reduced by introducing the standard deviation.

This paper introduces the control method for the impact force based on feed-forward control considering the stiffness ellipse of a biped robot. Assuming dynamic motion, such as jumping and running, the jumping direction depends on the dynamics in response to the ground reaction force. Therefore, it is necessary to control the ground reaction force in order to stabilize the system. However, control of the ground reaction force involves a fundamental problem: feedback control does not work properly against the impact force owing to the control bandwidth limitation. This study introduces a method for the stiffness ellipse control utilizing three antagonist pairs of six pneumatic artificial muscles. When the tip of the biped robot makes contact with the ground, an external force is induced in the direction that has lower stiffness. By utilizing this property of the stiffness ellipse, it is possible to control the ground reaction force by feed-forward control, which is not dependent on the control bandwidth. The results of several experiments convince us that the relationship between the ground reaction force and the stiffness ellipse is almost linear, and the ground reaction force can be controlled with high reproducibility by adjusting the stiffness ellipse. It is possible to stabilize the system during dynamic movements because this method can control the impact force.

In this paper, a lightweight master interface for a surgical robot system is proposed. The position of the human hand is measured by a non-contact motion tracker, but the force feedback is realized by a handheld haptic device. A gripping force is displayed by a pneumatic bellows, which is smaller and more lightweight than electric motors. As for the slave manipulator, the gripper of a surgical forceps is actuated using a pneumatic cylinder. Force estimation is possible due to the back-drivability of the pneumatic actuator. Experimental results show the control performance of the haptic device and a constructed master-slave system.

Hydraulic actuators are superior to electric motors in term of power/weight ratio. In addition, recent technical progress of electro-hydraulic valves provides fast tracking performance. However, there is a problem of steady state errors because hydraulic actuators have dead zones around zero outputs. Disturbance observers are powerful tools to eliminate disturbances on robot systems. However, if there is a dead zone, disturbance observers generate residual oscillation. Feedback modulators cancel Coulomb friction by quantizing inputs. However, viscous friction, gravitational force, modeling errors, and the other disturbances cannot be compensated by feedback modulators. Therefore, in this paper, the dead zone is compensated by integrating disturbance observers and feedback modulators. The validity of the proposed method is experimentally confirmed.

This paper describes a technique to provide tactile sense to a desk by a simple mechanism. The tactile sense is provided to the whole body of the desk by fixing a force sensor on each leg. One of the advantages of this technique is that the contact point can be estimated without any sensor device on the surface of the desk. The desk can be used as a kind of a tactile interface by utilizing the advantage. Some examples of application using the desk with computer graphics are introduced.

As a rehabilitation support system is a digital device, it has an ability to automatically record and store data. In this study, we utilized this feature to develop an operation interface. Operation interfaces can be compared on the basis of the recorded data. In the case of the stick-type operation interface, the operator can freely control the device by recognizing the direction and magnitude of the applied force. During the stick operation, an appropriate degree of rigidity for the stick is necessary to enable operations involving both fine and large movements.

A whole-body tactile sensing system using a force sensor with soft material at contact areas is proposed. The previous method proposed by the authors used rigid materials at contact areas. In terms of safety assurance, the material should have shock-absorbing characteristics. Hence, in this study, we investigate the use of an urethane sponge at contact areas and evaluate the performance of the contact point calculation. First, the relationship between the external force and the displacement of the urethane sponge is measured. Second, a compensation method for deformation of the soft material is proposed. Finally, the performance of our whole-body tactile sensing system is verified.

Working in ultimate environments (e.g. working at the disaster sites) is dangerous for humans. Hence it is important to introduce teleoperated robots at the extreme environments. Since there are many heavy works, hydraulic actuators of high output power are used more than electric actuators. Therefore, it is necessary to operate remote control robots driven by hydraulic actuators. Bilateral control is remote control with the sense of force. Because most of bilateral control systems have been studied mainly using electric actuators, bilateral control using hydraulic actuators is still a challenging issue. In this research, we show the interference of position and force controllers and degradation of control performance by scaling gain in the conventional method. The control performance of the conventional research is analyzed and is experimentally verified. In addition, in this research, we show the effect of modeling error of the previously proposed bilateral controller.

In exercise therapy, the training program will differ depending on the degree of disability. In order to gradually transition from passive exercise to exercises with more voluntary movements, it is important to reduce the assistance provided by the therapist in stages. Since stiffness control is the key factor of assistance adjustment in robotic movement rehabilitation, this paper focuses on stiffness control as a tool to adjust the assistance in stages. It is necessary to set a proper stiffness ellipse to perform assistance with directivity, especially in the case of active-assistive exercise. The performance of the proposed stiffness control for exercise therapy is validated through experimental results.

In this paper, a way to obtain bilateral control between different time resolution systems is proposed. Bilateral control enables communication of information regarding our tactile sense. A master (local) robot is manipulated by an operator and a slave (remote) robot tracks the position of the master while the reaction force of the slave is fed back to the master. Conventionally, there have been few studies about the difference of time resolution and feedback gains between master and slave systems. First, an extended Jacobian matrix is proposed to describe the dynamics of bilateral control taking time resolution difference into account. The fast (master) and slow (slave) systems are transformed to other fast and slow systems in modal space, which the dynamics of the position accordance and force regulation are described in. Then, a multirate controller is designed to couple the fast system with high feedback gain and the slow system with low feedback gain. The validity of the proposed method is confirmed by simulations and experiments.

It is a quite wondering fact that all land animals and humans have biarticular muscles although the extra biarticular mechanism seems unnecessary in engineering view. This research investigates the contribution of biarticular muscles in living organisms to the attitude control. A method of angular stiffness calculation is proposed and the maximum angular stiffness of each musculoskeletal model is estimated based on the method. It is shown that the model with biarticular muscles can set higher angular stiffness without increasing the total mass of actuators. Furtheremore, experimental results show that control bandwidth is extended due to the high stiffness property.

Whole-body force sensation with a simple configuration is desirable for robots in human environment. This paper deals with a force sensor-based whole-body force sensing technique of robots and expand it to arbitrary shape end-effectors. Previous techniques could not deal with a non-convex-shaped end-effector since multiple candidates of contact points exist for contact position identification. The problem was solved by explicitly using the property of an external force vector whose direction changes with respect to time. It was found that inertial force degraded the detection rate in cases in which vibration occurred in the end-effector, but it was possible to improve the rate by acquiring data over a longer period of time. It was empirically confirmed that accurate whole-body force sensation is feasible using a non-convex-shaped end-effector.

Haptic modality is one of the most important considerations in man-machine communication. The present paper introduces a personal identification technique based on force information of finger tracing that can be used for advanced man-machine communication systems. Calculation time is evaluated to determine whether the method is appropriate for real-time systems. Compared to other biometric methods, the proposed method has the advantage of high security due to the non-visible nature of force information. Experimental results demonstrate the validity of the proposed method.

High-speed cameras make it possible to detect luminance variations in images of a scene under AC illuminations. This paper proposes a method for estimating an image when AC illumination components are eliminated. If the eliminated illumination component is the source of specular reflection, the image produced only contains the diffuse reflection components of other illuminations. Removal of the specular reflection enhances the adaptability of a robot vision system to real environments. Experimental results show the advantage of the proposed method over using previous methods. The results also show that the method is applicable to moving objects while some errors are generated depending on the illumination frequency and the speed of the objects.

Recent development of motion control technology brings robots to our daily life. In the daily life, physical distance between humans and robots get very closer. This closed distance leads new issues in human robot interaction, such as safety issue and communication issue. The authors developed haptic sensing system called haptic armor. The sensory system can detect any contact on the robot surface and also it can distinguish the haptic command using contact information. The haptic command is the command which is delivered by physical contact to the robot. The present paper proposes the haptic commands for human robot interfaces. Furthermore, the design algorism of the haptic command is established in the paper.

Separation between specular reflection components and diffuse reflection components is an important subject in robotics and machine vision. Hence, the author's group has developed a method to separate a diffuse image from an image of high-speed video with specular reflection components. The method is based on an estimation technique based on luminance variation due to flicker of a light with an AC source. However, the method has a serious issue that it is limited for lights with an AC source. This study gets rid of the limitation by implementing a high-speed strobe light to the camera and produce flickers by artificial means. Experimental results show the validity of the proposed method.

Force sensors are a useful tool for robot adapting to human environments. However, these sensors are rarely used for commercial machines requiring safety since it is difficult to eliminate the possibility of failure. Although fault tolerance is an important issue, no critical method for general force measurement is proposed in the past. Hence, this paper proposes a fault tolerance measurement method using sensing devices with redundancy. The proposed method accomplishes both: fault detection without additional sensors; and force estimation during fault period. Furthermore, the fault detection handles many kinds of faults such as disconnection of wires, peeling of strain gauge, and so on. The validity of the proposed method is verified through some simulations and experiments.

This paper describes a method for command recognition by a haptic interface on a human support robot. Feature quantities are derived from the contact trajectory when an operator runs his finger across the haptic interface. The robot then recognizes a command based on these derived feature quantities from the symbolized contact trajectory. Conforming the robot motion to the symbolized contact trajectory allows for intuitive operation of the robot. The experimental results indicate that novice operators using this interface can obtain high command recognition rates.

Android robots that have a human-like appearance have been developed recently. The purpose of android robots is to realize realistic communication between human and robot by implementation of human-like behaviors. Then knowledge and techniques for generating human-like natural motions and behaviors with android robot are required. In face-to-face communication, we can observe that the speaker nods during his utterance, and we call this behavior "Speaker's nod". Thus, in this study, we tried to realize realistic "Speaker's nod" as a natural interpersonal behavior with receptionist robot SAYA that has human-like appearance. Therefore we experimentally investigated expression-timing, angle and angular velocity of the human "Speaker's nod", and we implemented "Speaker's nod" to SAYA based on the investigation results. Then we have verified that "Speaker's nod" at appropriate timing affected the humanity of SAYA, and it greatly enhanced emotional impression of human by subjective experiment.

A "muscle suit" that will provide muscular support for the paralyzed or those otherwise unable to move unaided is being developed as a wearable robot. The muscle suit consists of a mechanical armor-type frame and McKibben artificial muscle. Using a new link mechanism for the shoulder joint which consists of two half-circle links with four universal joints in total mounted at both ends of each link, all motion for the upper limb has been realized. Applying the muscle suit to non-healthy people such as elderly people and/or disable one requires very high-level safety and usability, from the practical point of view. We then in the first place for the practical use, apply the muscle suit to the manual worker. In this paper, we introduce how we can apply the muscle suit for the manual worker.

The goal of this study is to investigate the intelligence necessary for a machine such as a human-friendly robot working in an environment where humans and robots coexist. It seems to be required to manufacture a human-friendly robot that can communicate with human beings as a test bed or platform. Since face and facial expressions are crucial factors for communication, we have been developing a face robot which has a human-like face and can express facial expressions similar to human being. The important factors of the face robot for coexistence and communication with humans are: 1. human size and compactness; and 2. ease of control. While we have used air cylinders with pressurized air for the first-generation of face robot, to accomplish these two factors, we decided to use a SMA (shaped memory alloy) actuator driven by electricity and built the second-generation of face robot. Although we succeeded in making a human size face robot, it came to clear that the SMA does not have enough power and speed. To overcome these issues, we developed an air-tube-actuator driven by pressurized air and manufactured the third-generation of face robot. In the paper, we show how to build the third-generation of face robot and its basic ability.