EEG signal is one of the biological signals that can be useful to control wearable robotic devices, according to the human motion intention. However, the estimation of the user's motion intention from EEG signals is not easy. The user's motion intention might not be estimated when the user does not concentrate on the control of the robot and is distracted by other things. In this paper, a neural network based real-time estimation method is proposed to detect human motion intention using EEG signals.

抄録全体を表示

This paper describes a tactile sensor system that is aimed to be equipped on a robot hand. The proposed system consists of four pressure-sensitive sensors and a plastic board. A board with four foots is placed on the sensors. Stick-like objects are our target, and we aim to estimate object type and its grasped direction. In the estimation phase, pre-generated two discriminative functions output the two classification results. The experimental results show that the proposed system has a value as a sensor suitable to be embedded in a palm of a robot hand.

抄録全体を表示

In this paper, we propose a self-localization method with sound source separation using pulse-compression for mobile robot. The proposed localization method utilizes the known sound sources as landmarks, which are detected by the microphone array. Pulse-compression was executed to enhance beacon sound and separate the direction-of-arrival (DOA) of each sound source. The proposed method was evaluated by computer simulation. In the simulation, the localization error of the proposed method was 51% smaller than the method which separates sound using band pass filter. From the result, the proposed method indicates the robustness under the disturbances. This result contributes to the use of this method in the realistic, noisy environment.

抄録全体を表示

This paper describe a novel method for automatic assembly. In this paper, convey chain system is proposed to realize effective and reliable assembling system. And, proposed system is shown its ability through several experiments. The success rate of assembly task is 85.0%. This is not enough but correctable success rate.

抄録全体を表示

A method of making desired rock sediment shapes on a truck bed is developed by designing target motion trajectories of a wheel loader bucket. The behavior of the rocks in the bucket and on the truck bed in loading operation is analyzed by camera images and simulations. By combining both behavior models obtained by the analyses, an algorithm of designing the bucket motion to make desired rock sediment shapes is proposed, and its effectiveness is verified by fundamental experiments.

抄録全体を表示

We have been developing a lunar subsurface excavation robot using peristaltic crawling based on an earthworm's locomotion. In existing robots, the excavated soil must be manually removed by humans. To overcome this limitation, we have been developing a mechanism that automatically removes the excavated soil. This paper presents our new automatic soil-release mechanism and demonstrates the performance of the mechanism using a prototype.

抄録全体を表示

We developed a one-legged robot with straight fiber-type artificial muscles and a magnetorheological brake for jumping. In this study, we produced a one-legged robot with 2-DOF, structured the jumping model considering landing and calculated the jumping height by simulation. The result of the simulation shows that the robot was able to jump safely.

抄録全体を表示

Tactile sensors are very important components in robots, not only in robot hands but also in mobile robots since the robot is always in contact with the terrain. This study describes a new tactile wheel for a small robot, which detects a wheel-terrain contact point by an infrared proximity sensor. In spite of the small wheel size, this tactile wheel is high-resolution because the contact point angle is measured by a potentiometer installed on the outside of the wheel. We plan to use this tactile wheel for self-localization of a mobile robot on an uneven terrain.

抄録全体を表示

In this paper, we proposed that roll movement realized by yaw and roll command combination method of AR.Drone 4 rotor helicopter for an automatic position stop. Since there is a large effect in the position control of the influence of self-generated airflow disturbances and unpredictable airflow effect, it is necessary to realize a stable flight in an unpredictable difficult air situation. Our aim of this study is to realize the stable position control in an indoor room by proposing yaw and roll command combination method of the 4 rotor drone. Above customization realize the helicopter AR.Drone automatic stop control within standard deviation 12.4 [cm] in 3.5×5×2.4 [m^3] room. Our proposed method found an effective use of the degree of freedom of the drone movement yaw and it will useful for hobby-class 4 rotor system's the aircraft position control under the unpredictable airflow condition.

抄録全体を表示

Assessment and maintenance of the swallowing function is important to a person's quality of life. Recently, the swallowing sound has been used to assess the swallowing function. However, start time, end time, and duration of swallowing were manually obtained by visual and auditory inspections in many previous studies. Their automated detection needed. Therefore, in this study, we developed a swallowing time measurement system that can automatically obtain those metrics from the swallowing sound. We then evaluated the detection performance, and confirmed that the developed system can correctly detect them 87.1% of the time.

抄録全体を表示

Dementia is one of the higher neurological dysfunctions that reduce the quality of life (QOL). External support can enhance conation and self-efficacy of dementia patients. A remote operation robot attracts our attention as an external support. It can intrigue dementia patients, but, it is impossible to enhance self-efficacy because there is no active involvement of dementia patients. We suggested a robot race game to enhance self-efficacy and developed a robot operated system that dementia patients could use for physical exercise. It was recommended to enhance the self-efficacy of dementia patients because they laughed operating the robot in this study.

抄録全体を表示

Hemiplegia in patients causes a foot drop that cannot be grounded at the heel; therefore, the heel rocker does not function adequately. Currently, although the ankle-foot orthosis is used in these patients, its dorsiflexion assistance is not enough for the patients, and furthermore, it has not been adapted to the heel rocker support. In order to solve this problem, we developed an artificial muscle-driven high-dorsiflexion support RT (1st prototype). However, the responsiveness of the artificial muscle is too slow to support the heel rocker. Additionally, it cannot be adapted to the cadence. Therefore, we developed a high-dorsiflexion support RT (2nd prototype). The high-dorsiflexion support RT 2nd prototype is adaptable to cadence and incorporates the spring system to compensate and solve the problems of slow responsiveness.

抄録全体を表示

Mountain forests in Japan are steep with weak subgrades and many obstacles on the ground surface. The forest road density is also low (16 [m/ha]). This means that forest workers have to travel in forest areas with no roads while carrying loads. The authors have developed a movement and conveyance machine based on a crawler for the purpose of use in forest areas. This report proposes a crawler mechanism equipped with a new suspension mechanism and shows the effectiveness of the proposed mechanism in an actual trial.

抄録全体を表示

This paper presents a method for generating a whole-body squat motion for humanoid robot HRP-4 from human motion data. We convert joint angle trajectories into linear combination of base functions which are cubic B splines. In order to take into account the interactions with the environment and the stability conditions, we added appropriate kinematics constraints to the trajectory optimization to perform smoothing of whole-body motion. We investigated the validity of this method by using dynamical simulator and we succeeded in simulating squat motion with this method.

抄録全体を表示

Recently, the number of devices using the sensor technology to improve various sports techniques has increased. However, there are only a few devices that would support the improvement of training skills in basketball. In the research involving the development of a skill support device for basketball, i t is important to improve the free throw skills to acquire a physical posture of set-form. Therefore, we have developed a device for reforming the set-form by using auditory biofeedback. The proposed device measures the shoulder angle in real time and informs the ideal physical posture by a sound based on measured angle. According to our study for testing the efficacy of the device, the intervention group obtained 20% higher success rate than the control group.

抄録全体を表示

Hemiplegic patients cannot sufficiently weight shift to the affected side because of fear of falling. We aimed to invent and evaluate the Implicit Method that directs weight shifting to the affected side without it being noticed by the patient and reduces the fear of falling down. This method uses vibratory biofeedback to provide center of foot pressure displacement. A clinical study was conducted with 5 hemiplegic patients and the results showed that during walking, the stance phase of the affected leg increased after training in moderate paralytics. The results suggest that the Implicit Method can make hemiplegic patients stay on the affected leg while walking.

抄録全体を表示

In this paper we present an online unsupervised method based on clustering to find void-type defects in concrete structures using hammering. The dataset of sound samples is clustered in order to find the regular model for the hammering sound, which is assumed to be the non-defective sound model. The algorithm is fast and reliable enough to allow efficient diagnosis by running it each time a new sample is acquired.

抄録全体を表示

In order to enhance the usability, the silicon retractor attachable to suction pipe was developed. However, retracting force measurement is also preferable in order to avoid injuring brain tissue due to an application of unexpected large forces. With this in mind, in previous study, we developed force sensing system embedded in silicon retractor attachable to suction pipe. The developed system can provide three functions of suction, retracting, and retracting force measurement at the same time. However, the evaluation was limited to the cases when the developed silicon retractor contacts with flat surface. Investigating the performance evaluation in the situation close to real operation is necessary for the usage at real operations. Therefore, in this paper, we investigated whether the developed force sensing system can be valid at the various surfaces. Then, the developed system was considered to be valid to the application in real situation.

抄録全体を表示

Robot design is a subject of critical importance in a context in which robots are supposed to interact with humans. Head is particularly important, as it is the main part of the body that can convey emotions and other messages. In this paper we investigate the meaning conveyed by the use of coloured illumination. Through a survey, we obtained interesting findings which will be used for the development of the new head.

抄録全体を表示

This paper proposes a navigation method using rotating electromagnetic coils for a surgical intravascular microrobot in a pulsatile flow. A Proportional-Integral (PI) controller capable of dynamic tuning its proportional gain was developed. The microrobot was navigated to a specified branch of a two-branched vascular model, and the success rate was measured. The result suggests that the controller is robust and has potential for practical applications.

抄録全体を表示

We present acoustic event identification by integration of sound source separation and deep learning based on a convolutional neural network for extremely noisy acoustics signals captured with a 16 ch microphone array embedded in an Unmanned Aerial Vehicle (UAV).We showed that the proposed method can identify over 98% sound sources correctly for a 10 class classification task using 16 ch recorded sound data with a microphone array embedded in a quadrotor.

抄録全体を表示

This paper presents a method of reducing load friction in two degrees-of-freedom (2-DOF) transparent electrostatic induction actuator by using vibration-induced squeeze film effect. An experimental set-up was built to prove the concept. An overall 70% reduction in required driving voltage was obtained when the squeeze film is present.

抄録全体を表示

Tilted visual information induces body inclination on upright posture; however the effects of tilted visual information on control of lateral weight-shift posture or gait are unknown. We studied the effects of tilted visual information on the control of upright posture, lateral weight-shift posture, and gait in healthy adults. Three subjects participated in the experiment. Their visual information was tilted toward left using the head mount display (HMD), and their postures and motions were measured. Their body inclinations to the left were clearly observed under gait, whereas these inclinations were slightly observed under upright and lateral weight-shift posture. It was shown that the effect of tilted visual information on dynamic condition such as gait was greater than static condition such as control of upright posture.

抄録全体を表示

Virtual Reality (VR) technology can give users a sense of presence and an immersive experience by providing artificial stimuli for the user's senses. The purpose of this study is to develop a quantitative evaluation method that can evaluate system usability or efficacy with the aim of assessing systems that have force display functions. In this paper, we give an overview of the work and describe the results of our experience using an application that we developed for the study.

抄録全体を表示

Differential Visual Streams (DiVS) is a method to recognize common locations and quantify viewpoint differences between two image sequences, enabling robot self-localization along known paths (for which reference images exist) from images recorded by a single uncalibrated camera. It combines concepts from memory networks and convolutional networks, recent developments in machine learning with complementary features. Experiments show DiVS provides a sound basis for an image-based teach-replay navigation system working both indoors and outdoors, against variations in lighting and landscape composition.

抄録全体を表示

In recent years, various power assistive devices with electric, pneumatic and hydraulic drive have been developed. The advantages of pneumatically-driven system such as light-weight, high power-weight ratio and direct drive are suitable for the devices. However, the method to provide air power source is a critical issue in pneumatically-driven devices. To solve the problem, we propose and develop a pneumatic walking assistive system with a soft exoskeleton and a follower robot carrying an air compressor. The exoskeleton is actuated by pneumatic artificial rubber muscles (PARMs) and walking intention is detected from the pressure change in the PARMs. The follower robot follows the wearer with a tether interface. The effectiveness of the system is confirmed with some experiments.

抄録全体を表示

We focus on active vibration control of a flexible manipulator, and have implemented an adaptive notch filter based on Filtered-X LMS algorithm (FXLMS). FXLMS is commonly used for audio equipment in order to suppress a specific frequency bandwidth by applying the anti-phase wave. We applied it for suppressing resonance frequencies of the 2 DOF flexible manipulators, and investigated the vibration stabilization time between with/without the vibration control. As results, it took 8.35 [sec] to stabilize the vibration with no vibration control, while it took 0.75 [sec] with the proposed control. Therefore, the proposed method demonstrated its effectiveness for simple tasks.

抄録全体を表示

This paper evaluates breaking performance of an electrostatic surface induction actuator. The actuator is equipped with piezoelectric vibrators such that the friction between the slider and the stator electrodes can be dramatically reduced by squeeze-film effect. In such an actuator, the friction force can be changed by turning on and off the vibrator. The friction change can be utilized for high-performance slider motion control; for example, friction can be increased by switching off the vibrator when the slider needs to stop. In this paper, we evaluated how fast the slider can stop in several conditions. The result clearly shows the effect of friction change in breaking performance of the actuator.

抄録全体を表示

This paper presents the method of motion control for welfare mobile robot to realize traveling support of wheelchair user. The proposed welfare robot has both grasping function and mobile function. It is shown that by using active caster opposed dual-wheel as transportation device, wheelchair can move at arbitrary rotational radius and translatory velocity in keeping only grasping force control without positioning control by arms.

抄録全体を表示

This paper presents a new design of braided angle soft actuator for flexible bronchoscopy tip, which combines twisting and bending motions of soft actuator. This study is focused on the design, simulation and manufacture stages. The soft actuator was designed to fit with the anatomy of human bronchial tree for inspection purpose. The simulation of the design was done to verify the best structure for producing the functional requirement. The fabrication of the soft actuator utilized a molding system. From the gradient value of linear regression analysis, it demonstrated that the proposed soft-tip is capable of moving close to the functional requirement after standard pressure driving experiment was executed.

抄録全体を表示

This paper presents the System Identification (SI) method for modeling of thin soft actuator and its force control using PID-PSO. System identification is used to obtain the mathematical model (transfer function) from the measured experimental data using the SI procedure. Auto Regressive with Exogenous Input (ARX) model is chosen as model structure of the system. The result from SI model shows linear discrete model to obtain a discrete transfer function for the soft actuator system. Next, PSO-PID controller was proposed for the force control of the actuator. Validation was made between simulation and experimental data for the force control. Results show that the developed model represents the actual system by giving same characteristics in the force control analysis.

抄録全体を表示

In this paper, we analyzed spatial perception ability based on human eyesight for teleoperation. Recognizing the space relationship between different objects by human eyesight from a two dimensional monitor quickly and correctly is a hard task. On the basis of the analysis about visual perception error, we make some assumptions about the alignment of multi monitors. The reasons which cause visual perception error were then investigated using our virtual reality simulator,and experiments were implemented to verify our assumptions. Results show when objects were fixed, continuously moving the viewpoint around the fixed objects, the position of viewpoint obviously affected spatial perception accuracy of human subjects. We also found that subjects tend to estimate object position deeper than its real position.

抄録全体を表示

Sense of Agency (SoA) refers to the experience of controlling external events. SoA is considered to have predictive and postdictive aspect. Recent studies have investigated the relationship between SoA and event-related potentials, but these studies focused only on the postdictive aspect, while the brain activities related to the predictive aspect of SoA remain unclear. In the present study, we focused on readiness potential (RP) and examined the influence of prediction of SoA on RP. In the experiment, participants pressed a key to trigger a tone, and the event-related potentials before and after the key-press was recorded. In the normal predictive condition, all the tones were presented after key-press, thus participants could predict that their key-press would probably cause a tone. In contrast, in the low predictive condition, one third of the tones were presented before the key-presses, thus the causability of key-presses on tones were doubtful. The results indicated earlier RP in the normal predictive condition relative to the low predictive condition, although the difference did not reach significant level due to the small sample size. We will increase participants to confirm the difference in RP between the two predictive conditions in future work. We believe that our work provides important knowledge for the understanding of the neural basis of SoA.

抄録全体を表示

In this article we compare two methods of direction-giving: maps versus Microsoft's hyperlapse video at 4x speed. To evaluate the merit of each method we measure various user experience indicators. The results show that hyperlapse is effective as it reduces the median time required to reach a location by 30%. We also show that the time a user spends trying to understand the directions is a strong indicator of risk of getting lost.

抄録全体を表示

With the development of Internet of Things and wearable devices, human motion analysis in daily-living environment has attracted more attention. Accelerometers have been widely used in the quantitative analysis of human motion. However, the outputs of accelerometers include both the acceleration from gravity and the human motion accelerations as well as acceleration from noises. Consequently the accelerations of body segments of interest cannot be measured directly. In this paper, a wireless wearable sensor with a triaxial accelerometer, a triaxial gyroscope and a triaxial magnetic sensor was designed, and a new method by integrating a three-order low-pass filter and a Kalman filter was proposed to estimate the sensor orientation. This system is able to extract the acceleration only due to the body motion with a higher accuracy. It also can transmit sensor data to a personal computer using a ZigBee wireless module. The new fusion algorithm makes the motion sensor system suitable to various cases for analyzing the human daily activities.

抄録全体を表示