In this paper, we propose a novel depth upsampling method by residual interpolation (RI) that uses both a low-resolution depth map and a high-resolution intensity image. Our method is an application of the RI to depth upsampling, where the upsampling is performed in a residual domain following its success in the field of image demosaicking. Experimental results demonstrate that our method preserves the sharpness of depth discontinuities and outperforms existing well-known methods such as joint bilateral upsampling and guided filtering.
In recent years, the stroke patient is in a tendency to increase. Many of them have hemiplegia as sequela. It was proved that the paralyzed function could be restored for rehabilitation. Therefore, rehabilitation is provided by various methods now. In this study, we developed the rehabilitation support system for hemiplegia patient. A method of the rehabilitation include what's called mirror therapy. This mirror therapy system gives mirror movement and tactile sense information to hemiplegia patient. In this paper, we show overview of system and result of evaluation experiment.
For hyperspectral imaging, we proposed an edge-preserving filter, named hyperspectral Gaussian filtering, and its separable implementation for accelerating the proposed filter. Experimental results show that the acceleration has superior performance to the brute-force implementation and the other state-of-the-art methods in denosing. Also, we successfully achieves 70-times speedup with the acceleration.
The purpose of this study is to analyze the effects on the insertion path of an ultrafine needle injected into a tissue under the pressing force of an ultrasonic probe. We developed a hypothesis that the insertion path is affected by the stress field and the insertion angle that occurs with the deformation of the tissue. To examine the effect of each factor, we compared the deflection of the needle caused by both the stress field and the insertion angle with that caused only by the insertion angle. The results revealed that the insertion angle is a major cause of variation in insertion path. Furthermore, it is possible to reduce the variation in the insertion path using a guide to restrict the insertion angle with the deformation of the tissue.
This paper presents a novel sports training system. We developed a visual real-time feedback system that directly transmits a bird's eye view from a drone camera to a player's eyes. Individual physical ability, experience and intuition is important; however, in team sports such as baseball, football and volleyball, players need to perform spatial cognition such that they immediately understand where they, their teammates and their opponents are on the field as if they could see it from a bird's eye view. The objective is to validate the feasibility of the bird's eye view feedback system. In an experiment, nine participants moved and stood between two experimenters as a cut-off man with or without the system. Using the system produced significantly less error than with no system (p < 0.05). In conclusion, this bird's eye view feedback system can improve human spatial cognition.
In this study, we detected multicolor luminous landmarks by omni-directional camera installed to the mobile robot. Then, we modify the route of the robot by detected multiple landmarks for self-localization. In this paper, we propose selection method of three landmarks based on standard deviation of predicted value of self-localization, and we verify effectiveness of proposed method by the experiment.
We have build a micromouse type robot that has one Raspberry Pi, which is a single board computer. Since several major operating systems work on the computer, we can use various kinds of open source for this micromouse. This property contributes to robotics research and education based on recent open source movement. On the other hand, it should overcome the problem of interruption. We choose a structure that prevents running time of actuators becoming derailed by interruptions.
We propose a wearable line-of-sight detection system that utilizes micro-fabricated transparent optical sensors on eyeglasses. These sensors can detect the difference of reflection light intensity between white of eye and pupil. Up until now, we successfully have detected line-of-sight which based on 4 directions (up, down, right, left) and blink by using the practical application of our system. Our system showed experimentally that the error is as small as 1.5° in rotational angle of the eyeball. In order to develop practical application for our device, we improved our system to detect more precise eye's condition: open eye, blink, close eye. Finally, we devised an algorithm which process the information in these 3 conditions into eye state.
Recently importance of elderly nursing care is increasing. The nursing care can be decomposed into tasks which affect the elderly people directly and tasks which affects the elderly people indirectly. "Project to promote the development and introduction of robotic devices for nursing care" which was launched by the Ministry of Economy, Trade and Industry (METI) aims to support the former one. It is important to evaluate the effect of the devices in real nursing facilities. The paper presents future prospects to evaluate the robotic devices for nursing care. The first one is subjective evaluation. The authors have developed a framework for sharing data of business processes and events. The framework provides subjective information to evaluate the robotic devices. The second one is objective evaluation. The authors are investigating methodologies to estimate the absolute value of muscle force from myogenic potential. The estimated force will help objective evaluation of the robotic devices.
This paper proposed a force rendering method for palpation motions based on the haptic enhanced reality method. In this approach, the force from a haptic device is overlapped with that of an elastic surgical phantom to provide cost-effective force rendering. The experimental results showed that the proposed method can render heart beating on the surgical phantom.
This paper introduces an exoskeleton for reduction in a metabolic cost during running. The exoskeleton aims to reduce load of runner's vertical oscillation of center of mass. The exoskeleton consists of a leaf spring, a slider and a brake unit for holding and releasing the slider of each leg. The exoskeleton holds the slider during a stance phase and releases the slider during a swing phase. For assisting the oscillation, the leaf spring stores energy during an initial stance phase and restore the energy during a terminal stance phase. A heart rate of a subject shows assistance effect of the exoskeleton through running.
A novel hand-held robotic forceps for the laparoscopic surgery is developed. Intuitive operation is realized by an active transformation from the surgeon's wrist-rotation to the tip-rotation using an IMU and lightweight pneumatic actuators. The position control performance was evaluated and the in-vivo experiment was conducted to confirm the ability of suturing.
Achieving high control accuracy is a key issue for a surgical intravascular microrobot. This paper discusses the identification and optimization of the gains of a PI controller designed for the control of electromagnetic field for operating a 1 mm diameter microrobot. The microrobot was successfully controlled with a higher accuracy after tuning the control parameters.
A synthetic fiber rope which has lightweight, high tensile strength and flexibility is receiving a lot of attention as the replacement for a stainless steel wire rope. This paper takes note of durability of repetitive bending. We performed an experiment with two synthetic fiber ropes and one stainless steel wire rope in conformity with Japanese Industrial Standards. As a result, the stainless steel wire rope and one synthetic fiber rope which has high resistance to frictional wear didn't get loss of tensile strength. However, repeated bending causes the tensile strength deterioration of the other synthetic fiber rope which is weak in frictional wear.
For remote operation of robots in dangerous situations such as disaster sites, visual information affect the operational efficiency to a great extent. This paper presents a system which generates images as if users saw through obstacles by integrating images from two RGB-D sensors attached to a robot. The system is designed for use in more general situation than previous works.
This article presents feature extraction with deep learning for in-hand manipulation. It is important that robot hands can manipulate different sized and shaped objects. In order to generate versatile manipulation with such objects, we used deep learning to extract critical information from object manipulating motions.
This article describes our latest stationary underwater sensor node configuration that is used to construct an underwater monitoring sensor network (UMSN) to support the coral ecological system survey activities, and also describes about study of the required functions of a data browsing system to monitor information of underwater environmental conditions gathered by UMSN.
Home Oxygen Therapy (HOT) patients need portable oxygen equipment when they go out. However, the currently-used oxygen equipment is heavy enough to discourage HOT patients from their going out. Therefore we try to develop the power assist robotic cart which supports HOT patients' going out. In this paper we describe escort type cart development which moves next to the users. We propose new operation interface device which has high reliability and capability of easy-understandable operation, and then we verify its validity by dynamical simulation applying Open Dynamics Engine.
This paper proposes a wire-driven supernumerary robotic limb (SRL) that can work closely with a person like an extra arm. The SRL is attached to their body and assists the person in doing activities for daily life. Wire-driven mechanism was adopted for the SRL to secure the safety of the user and make the weight light. The authors designed and built a prototype of the wire-driven SRL and conducted experiments to confirm the feasibility of posture coordination by wires.
Many robots have already been released as research and development platforms. However, there are few robots that satisfy the requirements (such as size, function, and environmental safety) for domestic human support robots aim at. Our mission at TOYOTA is to develop partner robots that provide the freedom of mobility to all people and improve the quality of life (QOL) of all people. Such human support robots can move around the house, keep watch over family members, and fetch objects. Robots meeting these goals have three key features: (1) compact size, (2) safety and durability, (3) ease of use. We will give a status report on Human Support Robot (HSR), an example of such research and development platforms, and HSR Developers' Community.
We have proposed a control method of multiple teleoperated robots with robot service network protocol (RSNP). A shared map has been developed to perform a collaborative task efficiently. Owing to the map that shows the positions of robots and planned path, the operators controlled the robots efficiently in an exploration experiment. We describe the teleoperated robot system using a shared map.
Thin pneumatic artificial muscles are expected to be applied to make active fabric products like clothes and orthoses using their thin features. It is known that the peculiarities of cloth differ according to the conditions of weaving and knitting. In this paper, a preliminary experiment of measuring characteristics' differences in 4 patterns of textile woven (plain weaving) by thin artificial muscle and wool was conducted. The result indicated that the textile woven by the smaller (2 [mm], 7 [mm]) distance of weft showed decrease in max force output and shrinkage ratio compared to the none-woven actuator. On the other hand, textile woven by sparse pattern (15 [mm]) showed almost resemble characteristics of the none-woven muscles.
In unmanned construction, operators watch multi views to know situations at a disaster site. However, such a vision system requires much cognitive load of the operators to determine monitors and its parts which they should watch according to situation. In order to decrease cognitive load, we developed a system to provide appropriate monitors and its parts by inducing visual attention using augmented reality.
Some tools such as a compass or a lever contact at multiple points with environment. In the case of a robot using such tools, the individual contact force has to be estimated for dexterous control. This paper proposes a method for estimating individual contact force when two contact points exist. The validity of the method is confirmed by a simulation that a manipulator draws a circle with a compass.
Recently, various underwater vehicles, which are the useful tool to progress global important issues in relation strongly to the ocean, have been developed. Some of them are used to survey the condition of seafloor. They are equipped with some kinds of optical sensors. They take images of seafloor in order to clearly find significant points and measure the geometric information of them. One of methods to realize that is a stereo vision technology. In this paper, an underwater stereo vision system to visualize deep-Seabed three-dimensionally is described.
Omnidirectional cameras are widely used for robot inspection for their wild fields of view. However, the robot body will always be included in the view, causing occlusions. This paper deals with one such example of occlusion and proposes an inpainting based solution to remove it. Our method could generate a clean video automatically, without the need for a manually given mask. We propose an approximate shape fitting method combined with color information to generate the mask of robot body occlusion and followed by video inpainting. In experiments, the effectiveness of our proposed method is demonstrated by successfully removing robot body occlusions in omnidirectional videos.
Omnidirectional cameras are often used with robots to provide an immersive view of the surroundings. However, robots often have unstable motion and undergo rotation. In this work, we formulate a method to stabilize the viewpoint of Omnidirectional videos by removing the rotation using dense optical flow fields. The method works by first projecting each Omnidirectional video frame on a unit sphere, measuring the optical flow at every point on the sphere, and finding the direction that minimizes its rotational component in a frame by frame manner. The Omnidirectional video is de-rotated and a 'rotation-less, translation-only' viewpoint is generated. Such a technique is well suited to work in any environment, even with sparse texture or repeating patterns where feature correspondence based methods may fail.
The aim of this research is that a robot comforts a human to improve his/her mental state. It is known that human's facial expression is one of the triggers of arousing emotion. In this paper, as a means to comfort a human, we propose a method of predicting what kind of facial expression will appear. With this method, a robot selects an action which is expected to elicit a specific facial expression, based on the result of predicting human's facial expression. We performed experiments of eliciting a specific facial expression with Mof-mof robot. Mof-mof robot selected some actions which are expected to elicit designated facial expressions such as 'happy,' 'surprised,' 'angry' and 'sad' from four subjects. Some subjects' facial expressions seemed to be elicited by robot's action. Results of the experiments show that easiness of eliciting a specific facial expression seems to be affected by the type of the facial expression and human's will.
In this paper, for automation of hammering test, a methodology of constructing the defect detector that can identify defective locations is proposed. Particularly, defectiveness of material is indicated as a score, and the condition of the material is visualized by associating the score and the inspected position.
In this paper, we show a study on a planar non-constant curvature model for the continuum manipulator with rubber skin layer including pulling-wire mechanism. In order to derive the condition for calculating the shape of the manipulator, we make the assumption that a uniform distributed force per unit length is applied to the manipulator along the tangential direction of the manipulator backbone due to the friction between the pulling wire and the surrounding rubber skin layer.
In order to provide safe and precise autonomous driving for vehicles, a robust road lane detection and classification is needed. This research presents an image processing method based on Hough Transform, line filters and vanishing point, extracting road features parameters chronologically for estimating occluded lanes. Once the lanes are detected, they are classified in solid or dashed line. This method improved road lane detection and classification comparing to existing methods.
Lower back pain has become a serious problem in many workplaces. It is caused by lifting of heavy objects and adopting bad postures such as crouching. We have developed a power-assist suit that effectively eases the lifting-up motion. In this study, we evaluate the effectiveness of the assist suit in crouching positions. To this end, we selected two motion patterns of the assist suit and verified their effectiveness using a force sensor, motion analysis, and electromyography.
In recent years, robotic surgery has been widely performed in clinical practice, therefore the surgeons are becoming more and more familiarized with surgical robotic tools. Although the strong needs at clinical side, laparoscopy, one of the most performed minimally invasive surgery, is not widely performed using robotic system. To develop a further compact, lightweight, versatile surgical robotic system for laparoscopy, we designed novel robotic forceps developed based on Soft Robotics within the diameter of 6 mm. In this paper, we present concept, mechanical design and preliminary assessment of the developed prototype.
Master-slave surgical robots often employ a motion scaling function, and the surgeon's hand motion is scaled down with a motion scaling ratio and replicated by slave robot(s); however, the operation time can be prolonged if the task requires a smaller motion scaling ratio with a higher accuracy. In this paper, we propose the dynamic control of the motion scaling ratio according to the velocity of the master manipulator(s). Four control modes for controlling the motion scaling ratio were implemented in a master-slave microsurgical robotic system for neurosurgery, and the performance was evaluated by six subjects and compared. The results demonstrated that dynamic motion scaling has the potential for more efficient and comfortable manipulation.
Internal forward models provide predictions of movement consequences for voluntary motor control, and support adaptive behaviors to novel movement perturbations. We here addressed the question of what type of errors drives adaptation in internal models in a behavioral experiment. Three subjects adapted to visual rotation when sensory (a cursor position) or reinforcement (hit or miss) feedback were provided. The extent to which internal model adapted was assessed by the degree of online corrections to a new target location that unpredictably jumped at movement onset. The movements were corrected to opposite directions in the two conditions, indicating that the internal model of hand position did and did not adapt, respectively.
Drawbacks of conventional colonoscopy include the risk of accidental injury and a dependence on the operator's skill. This study presents a novel minimally invasive self-propelled colonoscope robot. We developed an active-bending module, made with pneumatic parallel links with three degrees of freedom, and assembled in a robot (WQE-5) propelled by a "party horn" mechanism with a thrust generator. We confirm the increase in the passing speed in the curved section of a colon model, achieved by the winding of WQE-5.
This work has been set out to improve poor accuracy and repeatability of passive-release by considering angle of a mechanical pickup approach between the endeffctors and a single micro-object. Due to strong adhesion force in microworld, the accurate and repeatable pick-and-release/place for micro-objects are required as a long-standing challenge in micro-manipulation. Moreover, due to nallow depth-of-field, microscopic 3Destimation is difficult. Thus, the system employs our developed depth estimation method on the microscope. The experimental results show the performance of the proposed pickup-and-place method.
Ladders are used in various places and for a variety of applications because ladders do not need to be installed in large spaces or specific structures. However, humans carrying heavy loads when using a ladder are subject to fatigue and the risk of falling. In this study, we proposed a robot that is capable of climbing various types of ladders using a simple link mechanism.
In recent years, the elderly population has been steadily increasing. In particular, the population is aging more rapidly in Japan than other developed countries. Surgical procedures for the elderly carry substantial risk, and there is an immediate need for measures to reduce this risk. We have been developing a support system for liver surgery by using IT technology. In this study, we focus on a high-accuracy measurement method for ascertaining the knife tip position with multiple markers during surgery.
In this study, we realize optimization of wheel position which prevents falling over for the leg/wheel robots. The prevention of falling over is realized by minimizing a performance index using model predictive control (MPC). The performance index is maximizing the area of a supporting leg polygon and realizing the center of gravity (CoG) position which prevents falling over. In this paper, we show the effectiveness through simulation.
This paper explains a motion planning strategy that is not based on time but on "event", which can be explained as the path that the end-effector of robot tracks in our study. Arc-length parameterization is used in this paper to convert a time-based planning into an event-based planning. With our planning, it is possible to choose various time planning on the same given path so that tasks such as dynamic obstacle avoidance and performance test of motors become available. Simulation results about time planning implemented to a given path verifies the effectiveness of our event-based planning.
We evaluated non-contact pulse wave measurement using near-infrared (NIR) sensor for casual sensing of vital information. This sensor is mainly composed of infrared LED and photodiode for measuring the cyclic change of hemoglobin concentration inside human body. To achieve the detection of scattering light from the human body for stable and precise measurement, the reflected light from the skin surface was removed by collimator. The theoretical model for design the non-contact measurement system was proposed. This NIR sensor could measure the pulse wave away from human body. Measured data was compared to the commercial pulse wave sensor as benchmark. The accuracy of the measured data such as heart beat rate (HBR) depended on the distance from the finger.
In recent years, accidents wherein roads have collapsed because of aging sewer pipes have been widely reported. To prevent such accidents it is necessary to inspect sewer pipes. We have developed a sewer pipe inspection robot inspired by peristalsis, the method that earthworms use to move. This robot comprises seven units, six joints, and a head part with a camera for inspection. In this study, we conducted field tests using actual sewer pipes, and confirmed the functioning, usefulness and potential of the robot.
This paper addresses a hypothesis of estimating elasticity of flexor digitorum muscle and tendon by measuring angle of MP joint. For ease of measurement and estimation, we propose a simplified musculoskeletal model of stretching of muscle-tendon complex (MTC), and formulate relationship between joint angle and resultant fingertip force attributed to MTC stretching. In addition, relationship between joint angle and angular velocity of released finger is also approximated. These hypothesis and approximation are verified by measurement experiments.
Developing affordable and robust human-machine interfaces is key to improve the quality of life of handicapped and amputee users in developing countries. While most of the available technologies are robust, the cost is inaccessible to large portions of the population with low-income. In this paper, we propose a compact, robust and cost-effective interface for arm pose sensing and control of external devices based on the low-cost accelerometer, gyroscope and Xbee technologies. Experiments using a prototype showed the promising results to detect and classify the robust poses.
Safety system is necessary when using industrial machines, but most of them decrees usability. In this research, we focused on vision-based safety system to keep usability. Different from usual usage of vision devices, vision devices in safety system must be located at upper area such as celling. The obtained visual information doesn't have face and body information. As a first step, this paper investigated whether human can be detected from visual information without those appearance features of human.
This paper introduces the development of a musculoskeletal model of swallowing to estimate muscle activities during swallowing including complex neuromuscular coordination. Individual muscle force is estimated by the inverse dynamics computation and optimization.
In this paper, development of a miniature size range image sensor is presented. The sensor is designed for a robot hand by using an endoscopy camera and a multi-slit laser projector, which can obtain range images. The improvements over the previous sensor are size, accessibility and measurement range in close distance. The dimension of the sensor is 16×40×53 mm and its weight is about 28 g. By using a universal webcam chip, the computer can capture the image data without the driver. The sensor can measure the distance from 50-150 mm. The effectiveness of the sensor is verified by experiments.
In this study, we focused on creating different scenarios and traffic conditions in a driving simulator to evaluate the differences in driving experience in conventional and autonomous vehicles. We conducted experiments using two groups of drivers and evaluated their subjective workload and preference for each driving method for different traffic conditions and scenarios.
In daily lives, humans do not always perform a single motion separately, but they adequately transit their motions to achieve complex movements. Although human sit-to-stand and walking motions are reported to be explained from the small number of modules (called synergy), it has not been revealed how the transit motion (sit-to-walk) is achieved. This study conducted an experiment to measure body trajectory, reaction force and muscle activation of one healthy participant during sit-to-stand, walking and sit-to-walk motions. Results showed that muscle activation during sit-to-walk motion could be explained from the muscle synergies of sit-to-stand and walking. Moreover, it was implied that each muscle synergy needed to be activated adaptively in order to generate momentum and successfully initiate the first step.