For a facial expression recognition model to retain its high robustness continuously along the time axis over a long period of time, the classifier created during the initial stage should evolve and gradually become adaptive over time. We propose a method for creating a facial expression recognition model that can offer such an adaptive learning capability.

View full abstract

In this study, we constructed an intuitive programming interface with a depth sensor for multi-fingered hands. We propose a method to control the hands just like by pinching and moving their fingers directly. This method was implemented for two types of two-fingered hands. From some experiments, we examined its advantages and problems to be solved.

View full abstract

This paper presents an analytical system for three-dimensional (3-D) reconstruction of underwater object using acoustic camera. We propose a novel analytical system which enables analysis of multiple acoustic images, leading to 3-D reconstruction of underwater object. The experimental results indicate that the proposed analytical system demonstrates outstanding 3-D measurement results.

View full abstract

The purpose of this paper is to investigate endpoint tracking accuracy of a redundant manipulator controlled in task-space feedback when the manipulator moves along external forces with satisfaction of specified tasks. The simultaneous realization of endpoint tracking and force adaptation was accomplished by redundant degrees of freedom (DOFs) and a controller consisting of the joint damping and the Jacobian transpose based endpoint errors feedback. The experimental results demonstrated that for endpoint position tracking of a 7-DOF manipulator the maximum tracking errors are within 0.21[mm] in adaptation compared to 0.17[mm] in no disturbance and the influence is significantly small (within 0.05[mm]).

View full abstract

In order to generate a path for obstacle avoidance, the Laplace potential method has been proposed. This method requires a great deal of calculat ion amount for generating potential. It is difficult to apply this method for the robot that operates in unknown environments because calculation costs for the method must be smaller. In this paper, the authors propose a new method which has small amount of calculation by combining the graph theory and the Laplace potential method. The simulat ion shows several results. First, the end point of the path generated by the proposed method is in the goal coordinate correctly. Second, the calculation time to generate path is smaller than the Laplace potential method. The authors remark that the success rate of path generation and the reduction amount of computational time are in the trade-off.

View full abstract

In this study, we developed a tactile display method that presents a realistic hitting sensation of a tennis shot using multiple vibrotactile stimuli. We have previously measured and approximated vibration waveforms at multiple points on a forearm during a tennis shot for displaying a realistic hitting sensation. However, the influence of swing velocity during impact on vibration parameters remains to be understood. Thus, we measured and approximated vibration waveforms in relation to swing velocity. The results revealed that swing velocity influences vibration amplitudes at the racket, wrist, and elbow during a tennis shot.

View full abstract

In this paper, we propose a method to predict motion over traversable obstacles for autonomous mobile robots based on 3D reconstruction and running information. Terrain images and vertical acceleration signal are collected and used. In order to estimate running features, function approximation methods are applied using image feature as input and running feature as output. The proposed method was evaluated by experiment of running on outdoor terrains.

View full abstract

In this paper we discuss a dynamic walking algorithm of a quadruped robot, especially improvement of its speed and stability. Previously we proposed the longitudinal acceleration intermittent trot gait, but with this algorithm a robot could only walk with constant target velocity and at higher velocity the robot fell over because of an initial acceleration. In order to solve this problem, the way of gradually increasing velocity in four legs support phase is introduced. The maximum acceleration and deceleration is limited to avoid complete tumbling. Additionally a swing leg trajectory is generated by using Ferguson/ Coons Curve. Walking experiments were conducted with proposed walking algorithm with the experimental quadruped robot TITAN-XIII. As a result the robot walked at a speed of 1.36 m/s stably.

View full abstract

Fiber reinforced Foamed Urethane (FFU) is a kind of artificial wood and is made of the plastic foamed urethane resin reinforced by long glass fibers. Due to excellent weather resistance, good machinability, lightweight and cost effectiveness, FFU is used as a synthetic sleeper. This paper proposes to use FFU as structural material for building a robot. Initial machining tests are also shown.

View full abstract

We propose a method of allocating multiple tasks among heterogeneous robots. The method is based on the theory of comparative advantage which is an economic theory. As an example, we apply this theory to carrying and exploring tasks performed by two types of robots.

View full abstract

This paper describes the development of a visualization system so-called haptic cutting board. The board can detect the external force vector and the position of the exerted force by utilizing the force/torque information from the 6-axis force sensor. The cutting board is supposed to be utilized to visualize force information during cooking.

View full abstract

To decrease human error in pericardiocentesis, we developed a pericardiocentesis assistance device that does not use a needle. The proposed device is able to aspirate the fluid in the pericardial space by cutting the pericardium with a cutter. To cut the pericardium, we designed a stress-type cutter and a shear-type cutter. We compared the two types, and found that the shear-type cutter can cut the mock pericardium.

View full abstract

In previous studies, we constructed a three-dimensional (3D) map using an RGB-D sensor and proposed a calibration method for this sensor. In this paper, after improving the calibration method for the RGB-D sensor, we examine the processing time and accuracy of a 3D map built with the corrected data.

View full abstract

We aim at analyzing human throwing motion with a sensor ball. So far, we developed the first prototype, which consists of a 9 axis motion sensor, a microchip computer, a data storage, a rechargeable battery, and, a set of plastic hemispherical cases. We also conducted a ball throwing experiment for the performance test. As results, we confirmed that the frequency analysis illustrated 4 motion states: loading, lifting, releasing, flying, and catching.

View full abstract

In this paper we describe a method to realize obstacle avoidance for a high DoF manipulator in order to allow the manipulator to operate in highly constrained and dynamic environments. Narrow spaces and new introduced constrains are considered. Conventional approaches for path planning do not utilize the previous motion experience or predefined data of path traversal thus the initialization is always done from a far less optimal starting point and then performs path finding using heuristics which are computationally expensive. We overcome this situation by enabling the robot to utilize a predefined set of data which is updated to include new constraints.

View full abstract

Neurological examination takes an important role in the clinical examination, and requires both deep medical knowledge and high clinical skills. Until now, several education methods for neurological examination have been proposed, such as lectures using videos, and skill trainings with manikins or simulated patients (SPs). However, complex physical skills, such as examination of the upper limbs, are difficult to obtain for inexperienced physicians both in lectures and in skill trainings. In this paper, we propose an arm robot named WKE-3(Waseda Kyotokagaku Elbow Robot No.3) for skill trainings of neurological examination, as a part of the whole body patient robot named WKP (Waseda Kyotokagaku Patient). A novel mechanism, which can simulate both active and passive movement, is implemented in the elbow joint to simulate various symptoms.

View full abstract

Hydraulic proportional valves generally suffer from their large size and heavy weight due to their complicated structures using large solenoids or motors. Therefore, they have some limitations of design or decrease mobility of hydraulic systems. In this paper, we present a small hydraulic flow control valve using particle excitation by a PZT vibrator. This valve can control the flow rate of water by the applied voltage to the PZT vibrator. The prototype of the proposed valve evaluated the relationship between the applied voltage, the impressed pressure, and the flow rate. This fundamental evaluation successfully indicated the flow-controllability of the proposed valve.

View full abstract

Designing optimal ZigBee networks is key for efficient low-cost communication of multi-agent sensors and mechatronics systems. In this paper, we report a novel approach to optimize ZigBee networks using the optimization of bundled routes. To the best of our knowledge, this is the first approach reported in the literature to find optimal ZigBee network topologies.

View full abstract

We propose a virtual probe system with use of image processing. By using the system, pre-scanned US images are shown on HMD by pointing a fingertip as a probe. The virtual probe is realized by an image processing algorithm. The algorithm detects locations of fingertip with colored glove and markers attached on the wrist and cubital fossa area automatically from images captured by a camera on the HMD. Feasibility of this system was confirmed by using US images of the forearm of a healthy person. The concept of this system was evaluated to be clinically useful by three nurses.

View full abstract

We developed a weeding robot that identifies and avoids registered plants. The robot extracts a single leaf image and identifies specific plants. It is difficult to extract a single leaf image from a camera image. We propose a method of extracting a single leaf image using multiple lights and a method of identifying plants by a single leaf image.

View full abstract

In this study, we propose an underwater robot which equips with a transformable frame to suit each situations of a mission. Specifically, the mission is partitioned into situations. Thereby we confirmed that the underwater robots required two movement characteristics when the situations of moving and monitoring. The situations are different by the independence of movement freedoms and the isotropic of thrust. Furthermore, we study the thruster arrangement for corresponding to the two situations. We discuss the usefulness of transformation me chanisms. From the above discussion, we extract the design requirements of the transformation mechanism, and developed the underwater robot which can change the movement characteristics by changing the thruster angle using only thrust power.

View full abstract

In this paper we present an investigation on body synergy analysis during robot assisted locomotion. We compare kinematic and muscle synergies of three hemiparetic persons walking with a wearable exoskeleton robot with two different control systems. The investigation reveals difference between the two systems, which can be used as a holistic criteria for evaluation of human-robot interaction in robot assisted locomotion.

View full abstract

In assembling large structures such as railway vehicles, escalators, and elevators, 3D measurement is important for verifying that parts are attached accurately. In this paper, a method to measure the 3D shape from the inside is proposed. Accurate cross-section shapes, obtained by the light-section method, are integrated in a new way with Structure from Motion using a laser. Further, an optical filter and a new block matching technique are introduced to avoid the influence of laser light on images while integrating the cross-section shapes. In a measurement experiment conducted inside a hall, the validity of the proposed method was confirmed.

View full abstract

Robot systems have recently been studied for real-world situations such as space exploration, under water inspection, and disaster response. In extreme environments, a robot system has a probability of fault. Therefore, considering fault tolerance is important for success of missions. In prior research, we proposed a distributed cooperative fault diagnosis method for internal components of robot system. This method consists of some diagnosis devices called diagnosers' which observe the state of a electrical component. Some diagnosers execute the diagnosis independently and in parallel, and we assume that they are connected using wireless communication. In this paper, we propose a technique which involves gathering the diagnosis results. Further we confirmed that this method can detect the faults of components in simplified fault situations by conducting computer simulations.

View full abstract

This paper describes a novel selection method of course direction for a mobile robot. We focus on the scene when a mobile robot cannot select an appropriate course direction by its own sensor information. The proposed method enables appropriate selection of course direction for a mobile robot by integrating measured data acquired by multiple mobile robots. Experimental result shows that the proposed method can decide appropriate course direction for a mobile robot.

View full abstract

In our previous work, we proposed a network-based activity-in-daily-life training system to enable a sighted trainer to train a visually impaired trainee to use various objects in a daily environment. The previous system had a problem that one sighted trainer was constrained to the training for one visually impaired trainee. During the training, the trainer was not able to train another trainee. In this report, we improve the system to allow a visually impaired trainee to retrain, by him/herself, what he/she learned from a sighted trainer. An experimental result of user study is shown.

View full abstract

This study describes the development of an omni-directional wall climbing robot with permanent magnets for the purpose of inspecting the walls of ships and gas tanks. The robot travels omnidirectionally and achieves stable movement through its large contact area. We equipped our traveling-wave-type omni-directional robot with a permanent magnetic adhesion mechanism and confirmed its wall-climbing ability. In this study, we analyze the magnetic field of the permanent magnetic adhesion mechanism by using finite element method. An improvement of the permanent magnetic adhesion mechanism based on the results of the analysis is foreseen.

View full abstract

A wearable force feedback device enables us to dynamically interact with a virtual object while moving in a wide virtual space. We have developed a wearable force feedback device with a pneumatic artificial muscle and a magnetorheological (MR) brake. These mechanical elements show quick response by feed-forward control and have high power density. In this paper, we tackled on weight saving of the device, and measured basic property again. Afterward, we evaluated whether operators could perceive the difference when rendered different magnitudes of frictional force or elastic force. The operators were rendered one of the five levels of force, and replied which level this force was.

View full abstract

This study aims to investigate how the kinematics of human arm (i.e. angle and angular velocity of elbow) at release timing (RT) affects on the outcome on the dartboard. For this aim, we detect the RT by tracking dart tail using motion capture. In addition, we examined EMG signals and compared three different muscles: triceps brachii (TB), biceps brachii (BB) and dorsal interossei (DI) between thumb and index finger to be clear expertizes. Results showed that there seems no any constraint between the two kinematic parameters with large variation in elbow angle at the RT in all beginners. Besides, in the EMG analysis, it is found that there was no co-activation of TB and BB, with shorter activation duration of the latter in the best performance subject whereas signals of DI illustrated the same trend for all players.

View full abstract

Every movement takes some time, but how such a movement duration is determined in the brain is still not well understood. The fact that humans exhibit task-dependent laws of movement duration suggests that there should be a computational principle that determines a movement duration. There are hitherto two distinct principles proposed. One is finite-horizon optimal control, in which a movement duration is predetermined and then movement is optimized during that duration. The other is infinite-horizon optimal control, in which movement duration is not predetermined and movement is optimized over infinite movement period. These models predict distinct, experimentally testable predictions regarding movement corrections against external perturbations. We performed a behavioral test with three subjects and report a preliminary result here. The movement duration issue is not settled yet.

View full abstract

To enable clean-up robots to vacuum cluttered floors, the robots should be able to return objects on the floors to the proper positions by pushing the objects. This paper proposes a computer vision based method that can detect the objects (clutter) and evaluate the result of the mechanical operation (pushing the objects). First, we construct a dense 3D map of a tidy situation of the room (after-clean-map). Next, we construct another dense 3D map of the cluttered situation of the same room (before-clean-map). Then, we construct the 3D point cloud of the cluttered situation by comparing the after-clean-map and the before-clean-map. Finally, we let a clean-up robot push the cluttered object and evaluate whether the object is successfully moved. Experimental results using real RGBD image sequences demonstrate promising results of our proposed method.

View full abstract

To decrease the risk of falls in older people, gait training robots are needed. However, compliance control, which is the usual method used in existing gait training robots, may cause patients to walk in an unnatural way. To solve this problem, the purpose of this study is to develop a gait training robot that applies torque to the knee only in the pre-swing phase. We investigated the influence on the minimum toe clearance in the swing phase from applied torque to the knee. We developed the robot and carried out an experiment in which a motor pulls a string embedded on the frame that the subjects wear, and applies torque in the pre-swing phase. The results demonstrated torque conditions wherein the minimum toe clearance remains high after applying the torque.

View full abstract

Previous researches reported that a lane change is the main cause of vehicle accidents. Therefore, inferring behaviors in other drivers and alarming to a driver are crucial tasks for safety. In this paper we propose a method to estimate lane change intentions in other drivers using Hidden Markov Models (HMMs). Comparing the proposed method with previous researches, we confirmed this method has higher accuracy to detect a lane change in other vehicles. And the method can recognize a slow lane change which is remained as a problem.

View full abstract

This study deals with a deskwork support system that assists human workers by delivering necessary objects to the workers using automatically moving trays. In this manuscript, we propose a new function that enables automatically putting away unnecessary objects using moving trays. Elapsed times from the delivery of objects are used to determine the time limit for putting away objects.

View full abstract

This study presents a joint mechanism that enables a manipulator to possess a multiple degrees of freedom with a single drive actuating unit. The joint called a Built-in-Clutch Joint (BCJ) is composed of an electromagnetic clutch, brake, and a belt pulley system. The joint is driven by a motor via the belt and pulleys when the clutch on the BCJ is engaged and the brake is disengaged. In addition, the joint angular velocity can be arbitrarily controlled by changing pulse width of input signal to the clutch. This study confirms the concept of the BCJ, and experimentally evaluates its mechanical design and performance based on the joint reachability and the transmission efficiency.

View full abstract

This paper presents a robotic system that can achieve automatic positioning of surgical instruments for vitreoretinal surgery. Visual servoing can assist the surgeon in performing vitreoretinal surgery by automation of certain difficult tasks such as drug injection into blood vessels. Our system allows the surgeon to select a desired target location for the surgical instrument on a microscopic image. Thereafter, the robot moves the surgical instrument to the target location automatically. This system consists of tracking of the instrument in the microscopic image and a robust visual servoing loop. This paper presents the results from the implementation of visual servoing loop in this system.

View full abstract

This paper studies generation of movement of a Limb-Type robot which climbs and goes through frame structure such as a ladder or Jungle-Gym. The proposed method for motion generation is based on Artificial Potential Method (APM) and physical constraints of a robot. Results of simulation show that the joint trajectory of limbs is generated to climb a ladder and to pass between rungs of the ladder. Movement maneuver of this study resolves a kind of complex motion planning of a Limb-Type robot equipped with multiple limbs.

View full abstract

During liver surgery, laparoscopic ultrasounds (LUS) require the visualization of blood vessels surrounding a liver tumor. However, this kind of imaging is difficult because of an insufficient positioning degrees-of-freedom (DOF) of LUS devices. We are currently developing multi-DOF LUS manipulators in order to realize autonomous functionality. First, we develop a method to detect the longitudinal section view of a blood vessel in acquired LUS images. Then, we successfully determine the longitudinal-section view by calculating the ratio of the blood-vessel area to the entire image area.

View full abstract

Static stretching is one of the most popular ways to decrease muscle tone. Muscle tone is evaluated by muscle hardness. To evaluate how muscle tone decreases during static stretching, we have proposed a method that involves using a wearable muscle hardness meter. In a previous study on the rectus femoris, we sometimes observed a muscle hardness increase during self-static stretching. We did not consider the degree of elongation of the muscle; however, this does actually affect muscle hardness. In this paper, we investigate the effect of joint angle on muscle hardness in self-static stretching of the gastrocnemius. We measured the muscle hardness of the medial head and the lateral head of the gastrocnemius and the dorsiflexion angle of the ankle joint during static stretching. In some cases, when the joint angle increased during static stretching, muscle hardness tended to increase. This result suggests that we should consider muscle elongation when evaluating muscle hardness during static stretching.

View full abstract

The authors propose a flexible shaft actuator whose slider has high-speed and long-stroke motion capability. However, the rolling resistance of rollers in the slider must be investigated to improve the motion performance. In this study, we identify the rolling resistance and optimize the roller design. First, possible factors causing rolling resistance are investigated. We found that the elastic hysteresis loss was the main factor to approximate the resistance. Finally, we optimized the roller radius using the experimental results.

View full abstract

This study proposes a concept of "Human-powered robotics," a design methodology for an intelligent system powered by human's motion. This paper illustrates the development of the several prototypes of the human-powered robot system.

View full abstract

Home Oxygen Therapy (HOT) patients always need to take the conveyance cart to carry oxygen equipment when they go out. However the equipment is heavy enough to discourage the patients from their going out. Therefore we try to develop a mobile robotic cart in order to support HOT patient's going out. by applying Force control and Inverted pendulum control. In this paper, we describe the control principle of the robotic cart by several simulations using Open Dynamics Engine (ODE), and we also show the effectiveness of our proposed control method.

View full abstract

Today, service robots have been gradually used to solve the problem of manpower shortage. One of these important uses is to provide multi-lingual service. How to understand users language and provide service effectively has always been an important issue. In this paper, we propose a multi-lingual service system based on the iBeacon network system. In this system, users with smartphones will save their personal information in a given APP which can exchange massage with the robot. Therefore the robots can get users information and provide services or push introduction message with appropriate language as soon as a person enters the iBeacon region. Also, collected information will be shared between devices, so that the robots can work cooperatively in the same area. We can expect this will provide a faster and more comfortable service than normal service robot system.

View full abstract

Pneumatic soft actuators are expected to be applied to various mechanisms in different fields, agriculture, medical, welfare etc. Axially Reinforced Pneumatic Artificial Muscle (ARPAM) has fibers lined along the length of its body which helps it to contract axially when pressurized. However, the fibers make the simulation process complex because of the difficulty in modeling and increase in computational load. In this research, a novel analysis method, realizing an instant simulation by introducing anisotropic material instead of fibers has been investigated.

View full abstract

Automation of agricultural machines has been required for labor saving to manage farms. The purpose of this research is to develop a vision-based guidance method for an autonomous vehicle in vineyard environments. In this paper, we propose a method of vanishing point detection focusing on similarity structure of vineyard environments. An advantage of the proposed method is not to be affected by illumination conditions. We applied the proposed method to 17 images and evaluated the usefulness by positional errors of a vanishing point.

View full abstract

The purpose of this paper was to develop an acoustic beacon signal for triangular surveying method in indoor location. The proposed method is based on the time of flight between the acoustic beacon and the single microphone. The plural amplitude modulated signals were applied as the beacon signal to reduce the effect of disturbing noises, and direct conversion receivers were used to cancel the difference of synchronization between the transmitter and the receiver. In this paper, we evaluated the number of modulated signals included in a beacon signal. The performance was evaluated by moving the microphone between a pair of acoustic beacons which were not synchronized to the receiver.

View full abstract

Gaussian Processes have been previously used to model wireless signals strengths and create location-signal strength mappings. Such mappings can be used for robot localization by computing the posterior probability distributions of robot's positions given signal strength measurements. This work proposes the use of a dual Monte Carlo Localization algorithm that uses such posterior distributions as perceptual likelihood. Our approach is assessed and compared with an argmax and a Monte Carlo Localization algorithm in terms of accuracy and computation time.

View full abstract

To progress the automated inspection for the inner wall of superannuated tunnel, a variable guide frame that maintained the inspection device has been developed. As this guide frame was composed of several VGT elements, this frame was able to fit the crass section of the tunnel shape and to evade the obstacle in the tunnel. The part 1 of paper, the principle of the guide frame and shape analysis are explained.

View full abstract

The running trajectory for bipedal robot is generated using numerically calculation method. Such method might limit the robot performance to accurately control its forward velocity. A new method is developed to determine the running trajectory based on linear equation that is developed based on walking trajectory and vertical hopping motion trajectory equations.

View full abstract