We have proposed a simple tactile nail chip as a wearable device that adjusts tactile sensitivity. In order to investigate the adjustment of the tactile sensitivity, it is necessary to evaluate the performance of the simple tactile nail chip. In this paper, we investigated the influence of time passage and wearing condition through evaluation test on roughness perception. The result indicated that the influence of time passage was very small, whereas the difference in wearing state influenced the tactile sensitivity change.

The aim of this study is to develop an efficient walking robot and walking support system. In this paper, we experimentally demonstrate a level walking of two-legged passive walker by traction force of weight. Moreover, we demonstrate specific cost of transport of the level walking.

The aim of this study is to understand the mechanism of human running. In this paper, first, we show characteristics of human running. Secondly, we propose a simple mathematical model of human running under air resistance. Finally, we analyze a fixed point and its stability for the mathematical model.

Aimed at improving energy efficiency of legged robots, flexible elements like springs are introduced to realize temporary energy storage. To achieve the excellent locomotion performance of animals, it is necessary to introduce the torsional spring at the revolute joint into the robot. This paper focused on a one-legged passive running. A passive running robot can run with no energy source without gravity. The trajectory motion is a limit cycle. Therefore, passive running may give us an important insight develop the biped running robots. In this paper, the stability of the fixed point of the one-legged passive running is demonstrated by simulation experiments.

In this paper, we develop a biped walker based on body dynamics and a phase oscillator. We have attempted to develop a 3-D dynamic biped with flat feet, ankle joint, and ankle spring, via a knee oscillation. In addition, we have applied the Owaki-Ishiguro method called ‘TEGOTAE’ function as ground contact information in the phase oscillator to the knee joint.

In this study, we aim to the human-like 3D walking using manual remote operation system with wire. This intervention is communication between the robot and the operator through force tactile and we can understand a walking state from a viewpoint different from data analysis. In the experimental surroundings, we realized 3D walking on level ground with no posture assistance by experimenters.

In this study, we aim to the human-like high-speed running that had high energy efficiency based on bouncing rod dynamics. We are intervening in the robot’s running by the remote manual operation system with wires. We realized human-like high-speed running on level ground with no posture assistance by experimenters.

We have developed a mobile robot that can easily change the number of wheels from two to six by detaching the driving modules.The robot can select the driving mechanism according to the environment.Confirmed that it is suitable in the three environment of grassland, gravel road and hill.In addition, Three-dimensional maps were generated using 3D LiDAR and NDT algorithm.

In general, the autonomous robots charge their batteries by using exclusive docking stations. However, these robots's operating range are limited around charging station's area and it takes cost each time to set up a new charging point. In previous researches, to solve those problems, we suggested a new charging method called “Plugin-docking system” that inserts charging plug into outlet directly with three-dimensional control with the Active-Caster Omnidirectional Robot. In this paper, we propose control methods in order to enable plugin-docking motion in various outlet cases. Also, we show the effectiveness of the proposed methods by experiments.

We experimentally tested the rubber hand illusion over a mirror where the rubber hand was seen in a mirror unlike typical experiments. The study was composed of three experiments. In the first experiment, we reconfirmed that the illusion was experienced when the participants stared the fake hand in a mirror. In the second experiment, the illusion was reported even with an incongruency between the actual and fake hand postures. Finally, in the third test, a smoke mirror was judged to be more effective than a clear one. These findings suggest that an obscure fake hand image in a mirror leads to a condition where the rubber hand illusion is readily experienced.

This study considers a newly developed, encounter type multi-finger haptic display. System is construct by two parts, one is the outer part using SPIDAR and inside is multi-finger haptic display. We have expanded the haptic system ability.

To address issues on conventional tactile sensors, such as low durability, sensitivity, and maintainability, we have proposed a flexible tactile sensor using a magnetorheological elastomer and a spiral inductor. To measure contact force distributions with an array sensor using several inductors, we should investigate interferences among inductors because our sensor works based on electromagnetic effect. In this study, we investigated how the inductance of the one inductor of a pair of inductors changes when another neighboring inductor is activated. The results demonstrated that the inductance increased when the neighboring inductor was activated, and that the amount of increase was reduced with a larger distance condition between the two inductors. We concluded that two inductors should be separated over 20 mm to maintain the sensor response curve obtained in a single activation condition.

We develop a system for psychological experiments to study the choice behavior of a human while he/she is pressing a button which has a vibration feedback. The system consists of a mechanical button with a piezoelectric diaphragm and a control unit. An experimenter can vary latency of the vibration feedback depending on the purpose of the experiment. Our evaluation showed that our system is capable of providing feedback in about three milliseconds at the best after detecting button press without chattering, which is considered to be sufficiently fast in order that a participant does not feel latency of the system. In our future work, we plan to first conduct a psychophysical experiment to study the human threshold for detecting latency of the vibration feedback. Then, we will conduct a psychological experiment to objectively and quantitatively evaluate the effect of the latency on the choice behavior.

Conventional optical tactile sensors using frame based cameras have difficulty in obtaining high temporal resolution since the time resolution of them depends on the frame rate of the camera used. We propose an optical distribution tactile sensor with high temporal and spatial resolution by using an event-based camera that outputs stream of events occurred by luminous change based on an asynchronous address-event method. The sensor measures the motion vector of its surface from the camera response of the checkerboard pattern. We measure when object is gripped and pressed using a robot hand.

This paper presents a design of a novel mobile robot based on a rimless wheel with a wobbling mass. We have shown that our rimless wheel can walk on level ground by effects of an up-and-down wobbling mass through numerical simulations. However, we do not experiment with the robot since we do not build the prototype. We have developed a mobile robot based on a rimless wheel with a wobbling mass. The robot consists of a rimless wheel and a torso with wobbling mass. We first show a design of our robot using 3D-CAD model, and confirm that our robot can walk on level ground through numerical simulations.

Bipedal walking locomotion is one of the characteristics of human behavior. Not only the lower body behavior but also the upper body behavior affects the walking characteristics. In this paper, we studied the effect of two types of trunk swinging motion on gait. One is the forward swing motion based on the bisecting hip mechanism. Another is the backward swing motion based on the inverted double pendulum model. A numerical bipedal robot model with an upper body controlled by these methods walked down a slope. We evaluated the gait characteristics, walking speed, step length, and step time, during stable one step.

The authors have been analyzing a motion generation of an underactuated locomotion robot with a quick return (QR) linkage through numerical simulation, and numerically showed that the robot is able to generate a forward motion by setting suitable parameters. In this paper, we outline the details of the developed experimental machine and environments, and report the basic experimental results. First, we describe the modeling and control law of the robot, and outline the experimental machine and environments. Then, we verify the validity of the numerical analysis results through experimental verification.

The authors proposed an underactuated locomotion robot utilizing effect of wobbling and sliding, and showed that stable and smooth sliding locomotion gaits on a slippery downhill can be generated through numerical and experimental investigations. They also proposed an underactuated locomotion robot with a quick return linkage aiming at generation of a forward motion on a slippery level surface. In this paper, for understanding the mechanisms of rotation on forward sliding, we introduce an underactuated locomotion robot with an actively-controlled inner reaction wheel. First, we introduce the mathematical model of the robot and develop the control law. Second, we numerically investigate the effects of the rotary derive of the reaction wheel on forward motion generation.

It is known that the walking is stabilized by using Constraining Sole Shape in passive dynamic walking biped. The reason is that an excessive body inclination is suppressed when the top of shoe touches on the ground. This effect is explained as the decrease of angular velocity caused by the change of the center of rotation from heel to the toe on a stance leg. The rate of decrease of angular velocity is calculated by using foot length and angle of ankle on walking model. By this, we analyzed this constraining Sole Shape effect by 2D numerical simulations on both sagittal plane and lateral plane. By these simulators, plots of stance leg angle and walking animation are derived. From the results, we confirmed stabilization by suppression of excessive body inclination on both planes.

In this research, we are going to verify how deep reinforcement learning and anomaly detection etc, which are drawing attention in the field of artificial intelligence, can be utilized. In particular, we accumulate know-how in application to real walking and running machines. In this paper, we report some results of preliminary experiments.

In this research, we aim to develop a road boundary estimation method by deep learning. By using deep learning, it is expected to be applicable to various road environments. Moreover, by performing time-series information integration, it is possible to deal with temporary mistakes of deep leaning.

This research aims to measure the position and velocity of moving objects using a stereo camera. In order to navigate a mobile robot in a dynamic environment while avoiding collision with moving objects and walking persons, finding objects and predicting their motion are needed. In the case of mobile robot, Unlike the detection of moving objects by using a fixed position camera, we need to take the ego-motion of the robot into account. In this paper, we propose a method which integrates stereo visual odometry, stereo matching, and optical flow to detect moving objects and estimate their motion from stereo images taken from a mobile robot.

Currently, convolutional neural networks (CNN) are widely used for object recognition, and it is common to use already trained network as a basis. However, the retraining cost of CNN to recognize a specific object is expensive in terms of dataset preparation and calculation cost. We propose instead to add a second CNN specialized on the recognition of the desired target. Additionally, a time-series filter is used to stabilize the detection of the target. To confirm the effectiveness of the proposed method, we used outdoor recorded movies and evaluated the accuracy of specific person recognition.

In recent years, Personal Mobility Vehicle (PMV) have been attracting much attention as the new future vehicle. In this paper, we propose a new control system which provides automatic driving function or driving support function to PMVs using a small sensor terminal named “Portable”. This system enables to add intelligent functions to various PMVs by attaching the Portable. In this paper we introduce two types of intelligent PMVs, standing type mobility and electric wheelchair, and show experimental results in indoor and outdoor environments including a theme park.

This paper describes a method for autonomous mobile robots following a human for supporting our lives closely. Robots have to recognize our behaviors and approach us for supporting. This paper introduces a robot system that follows a human with smartphone. Especially, this paper focuses on global following based on Wi-Fi radio field strength with many Wi-Fi access points installed in the environment. In this study, a Wi-Fi radio map is proposed for human following. The Wi-Fi radio intensity maps consists of Wi-Fi radio field strength and occupancy grid map of the robot. Positions of human with smartphone are decided in the Wi-Fi radio intensity map and the mobile robot travels to the positions autonomously. This paper shows some experimental results for calling robot and following human.

We propose an autonomous mobile robot system working outdoor with behaviors acquired by deep reinforcement learning for obstacle avoidance and topological maps for global path planning. In order to reduce learning time and acquire avoidance behaviors for dynamic obstacles, we developed our original simulator which can fast simulation including objects imitating dynamic pedestrians. Also, our proposed system aims to achieve navigation without occupancy grid maps. For example, global localization using RTK-GPS is combined with the learned behaviors for outdoor navigation.

Optical environment sensors such as microbeads modified with fluorescent dyes can be applied for minimally invasive measurement of cell aggregates and three-dimensional tissues. However, compensation for the fading effect of the fluorescent dye is a big issue. We fabricated hydrogel fluorescent sensors which can compensate the fading effect by fluorescent recovery. We conducted calibration of oxygen concentration measurement. Then, we examined stability and followability to environmental oxygen concentration.

This paper reports the measurement of the multimodal environmental condition such as pH and calcium ion concentration during the cell culture using fluorescence microsensors fabricated on the glass substrate. These fluorescence microsensors are coated with gelatin, which is used as the extracellular matrix, for the culture of the osteoblast cells on the gelatin surface. The fluorescence microsensors consist of hydrophilic photo-crosslinkable resin and three fluorescence indicators. Fura-2, FITC, and Rhodamine B are used for measurement of calcium ion concentration, pH, and temperature, respectively. The sensitivities of each fluorescence indicator are compensated by using fluorescence intensity of other fluorescence indicators. We measured local variation of calcium ion concentration and pH during conversion of octacalcium phosphate (OCP) to hydroxyapatite (HA) using fluorescence microsensors. We also demonstrated the culture of the osteoblast cell on the fluorescence microsensors.

This research group has established a technique to make a three-dimensional cell construct using only cells. The cell construct made uses cultured spheroids and the quality of the spheroids is important, it is necessary to provide reproducible spheroids. Therefore, it is necessary to morphology-evaluation spheroids. The quality of spheroids was evaluated by area, diameter, and circularity. However, the concentration distribution of the spheroids had not ever been visualized. This study aimed to grasp the concentration distribution visually using pictures of the spheroids and observed serial changes of spheroid's concentration distribution. It could grasp visually the spheroid's concentration distribution. We thought the spheroid serially condense cells and density was high at the same time. However it could confirm a part of low density. It is considered that the change which the cells move in the spheroid occur.

This research group has established an original technique to build three-dimensional Cells Construct in Regenerative Medicine without using Scaffold. Cells construct is made by the cell that was cultured spheroid. The quality of spheroids is important to make the assumed cell construct. Now, the quality of the spheroid depends on the skill of technical experts, and a lot of fluctuation factors exist, so it is required to eliminate fluctuation factors to minimum. Therefore, this research was aimed to develop automatic spheroid culture System and adjust the spheroid size. We used mesenchymal stem cell of rabbit for experiment. Observation of spheroid was conducted every 24 hours for 5 days. Results of comparing spheroids made by the developed system and technical experts, we were able to build spheroids with smaller diameter variations and higher circularity than technical experts.

For the donor scarcity problem, there are many Tissue Engineering studies. We fabricated toroidal cell structures inside a microfluidic device for tissue engineering applications. A photo-crosslinkable biodegradable hydrogel gelatin methacrylate (GelMA) was employed to encapsulate biological cells for assembling cell structures. stabilizeing the unstable GelMA manufacturing method,we review the manufacturing method and searched for a method that can be stavly produced. So changing the diakysis method, It was possible to prepare GelMA which is suitable as a cell culture medium. RLC-18 cells grow up on the micro structure. When RLC-18 cells grow up, GelMA is gradually reduced.

We propose skeletal muscle actuators with multiple degrees of freedom by using the flexible culture muscles. The actuators are composed of flexible tubes made with toroidal hydrogel and cultured skeletal muscle from myoblast cells. In this paper, the fabrication of cultured skeletal muscle was conducted as a first step of our research. The mouse myoblast cells with Matrigel was placed into a PDMS mold and pressed. It cultured to make thin skeletal muscles. The live/dead assay of cultured cells was conducted to check cell viability. The bonding experiment between cultured skeletal muscle and microstructure was conducted as a second step of our research. A toroidal structure made of PDMS or hydrogel was placed on the culture skeletal muscle. The cultured skeletal muscle and the toroidal structure were connected by placing mouse myoblast cells with Matrigel at the contact point between the culture skeletal muscle and the toroidal structure.

This paper proposes a method for generating a stable stealth walking gait without including the double-limb support phase on a frictionless road surface. First, we introduce a model of a planar underactuated rimless wheel with an upper body, and specify the target constraint condition of the total angular momentum around the ground contact point for maintaining the horizontal ground reaction force zero. Second, we derive an approximate analytical solution for the target initial state of the upper body by using the linearized equation of motion. Furthermore, we extend the results obtained to the nonlinear model, and numerically investigate the fundamental gait properties.

This paper proposes a method for achieving high-speed stealth walking of an underactuated biped without including the double-limb support phase, and discusses the effect of semicircular feet on the gait properties. First, we introduce a model of a planar underactuated biped with an upper body and semicircular feet, and derive the approximate target initial state of the upper body by using the linearized equation of motion. Second, we conduct numerical simulations to observe the typical stealth walking gaits, and analyze the changing tendency of the upper body motion with respect to the foot radius. Furthermore, we discuss the advantage of semicircular feet in walking on a slippery road surface from the viewpoint of the ground reaction force.

With the purpose of controlling the gait frequency of the robot locomotion, a novel and indirect method has been recently introduced using an active wobbling mass attached to limit cycle walkers. One of the strongest advantages of the method is the easiness of its implementation, prompting its applicability to a wide variety of locomotion systems. In this paper, we design an experimental machine for further verification of the numerical results. First, we introduce the typical modeling and control of this robot, a combined rimless wheel with an active wobbling mass. Second, we show the typical gait of it. Finally, we design an experimental machine for it.

Stable limit cycle walking is difficult to be achieved on slippery road surface due to the fact that the energy is consumed not only discretely at foot-tip collision, but also continuously at the grounding point. On the other hand, smoothly passive locomotion of the seed-like sliding robot is also difficult to be guaranteed without a periodic rhythm generator. To positively utilize the consumed energy by each other, here we connect them together by a rigid pole and conduct a preliminary analysis.

One characteristic of Passive Dynamic Walking is adaptive behavior. This adaptive behavior has been confirmed by simulation. Confirmation has been made in actual machine experiments with respect to the inclination angle as the external change. However, adaptive behavior against internal change has not been confirmed in actual machine experiments. Therefore, a viscous resistance generation mechanism which generates viscous resistance as an internal change is developed and experiments were conducted. In this paper, development of viscous resistance generation mechanism, operation verification, walking experiment are described.

Autonomous mobile robots need to move in an outdoor environment changing daily. Recently, simultaneous localization and mapping (SLAM) is often used for an autonomous mobile robot navigation. This method has good performance, but it takes time and effort to create accurate maps. However, considering the case of human, human do not need such accurate maps. Our previous study of human environmental recognition ability demonstrated that human uses scenery and road information. In this study, we propose an autonomous mobile navigation method with image processing based on human environmental recognition ability. This method uses GIST feature for localization, and SegNet for road detection. We validated our method in outdoor using a mobile robot. Experimental results demonstrate that the proposed method enables autonomous mobile navigation without accurate map.

To coexist with human, a robot has to avoid obstacles based on human-like flexible decision-making. In this article, we ran the robot manually in outdoor “Tsukuba challenge” course, took time-sequential images with the camera, and recorded the point cloud with LiDAR. By associating images with point clouds, we calculated the direction and velocity of obstacles on the screen. We recorded the angle and speed when a human operates a robot while looking at the screen to avoid the moving obstacle on a developed simulator. Using those, fuzzy rules to decide the moving direction and control speed at every moment to avoid obstacles were derived as follows: as the input variables, distance to obstacle (x₁), angle to obstacle (x₂), speed of obstacles (x₃), and the direction of movement of obstacle (x₄), are adopted. As the output variable, steering angle (y₁) and control speed (y₂) of robot is adopted. Based on fuzzy-neural networks method, two network of 4 inputs (x₁～x₄) and 1 output (y₁ or y₂) is prepared. Fuzzy rules are obtained so as that the network reproduces the trajectories of simulation experiment with minimum errors. The obtained fuzzy rules successfully produce controls which is similar to human ones.

In this paper, target tracking is one of the important functions for autonomous mobile robots. To realize this function, stereo cameras are often used. In a general stereo camera, a range image of the entire scene is acquired as precisely as possible, and the image is processed. On the other hand, in this paper, processing is performed by individual cameras according to the requirements, and the results are integrated to acquire three-dimensional (3D) information of only necessary objects. 3D tracking of an object is a specific task. Object detection and tracking are performed by individual cameras, and the results are integrated to acquire 3D tracking information. As a result, erroneous matching can be greatly suppressed, and optimum 3D information for necessary work can be obtained. The effectiveness of the proposed method is verified through target-tracking experiments.

In this research, we aim at autonomous navigation of mobile robot using braille blocks. We treat group of the braille block as a graph structure, let the point block as the node, and the line block as the edge. The robot moves to destination along the edge. We have created a point block detector and a line block detector. Both detectors process based on 3D point cloud data and images. The point block detector is constituted by a Fourier transform, and the line block detector is constituted by a Hough transform. In this paper, we evaluate these detectors at Shinjuku station. As a result, it was found that the detection rate of the line block detector is high in the block specified by JIS, and the false detection of the point block detector is large.

We study on acquisition of behavior to follow a specific person, while keeping a certain distance by using reinforcement learning. In the previous paper, we have already confirmed that the robot can acquires the behavior by using an angle of a pull string which connects the person and the robot as the reward of the learning. In this paper, to keep a certain distance, the stop action is added to the robot. And, the distance between the person and the robot is used for the reward in the learning phase. We confirm the effectiveness of the proposed method through an experiment.

Traveling in an urban environment, the autonomous mobile robot needs to recognize the surrounding environment in detail. As a conventional method, an obstacle detection method using the 3D point cloud given by LiDAR can be cited. The 3D point cloud given by LiDAR can acquire detailed position information, but it si difficult to recognize what the detected obstacle is. In recent years, object detection using deep learning has attracted attention, leaving detection accuracy exceeding the model base method. However, depth information using a stereo camera is unstable depending on the environment. In this research, we propose a real-time three-dimensional object position detection method by integrating object detection information obtained from a camera and three-dimensional information acquired from 3D-LiDAR.

A microfluidic channel is one of the best tools for the evaluation of red blood cell (RBC) deformability in a short time with just a tiny amount of sample blood. We have proposed a novel method to evaluate both viscosity and elasticity of a RBC in a microfluidic channel. For more precise evaluation of stress exerted on a RBC in a realistic situation, here we performed fluid simulation by finite element method. Calculating the flow profile and corresponding stress applied on the surface of a RBC, we obtained Young’s modulus and viscoelastic parameters through the image analysis of the cell flowing in the microchannel.

Red blood cells (RBCs) are subjected to intermittent deformation stress when they flow inside blood vessels throughout the body. We modeled such various deformation pattern using a microfluidic constriction channel mounted on an on-chip cell manipulation system, and measured viscoelastic responses of a RBC. Here we fabricated a microchannel with three narrow paths with different width, and observed shape recovery after the deformation through the paths under the feedback manipulation. Whereas the viscoelasticity of RBC is almost linear with respect to strain for a short loading time of tens of seconds, it becomes nonlinear for a long loading time of hundreds of seconds.

In recent years, as a powerful approach to elucidation of biological phenomena, "live imaging technology" which observes life phenomena of living cells and tissues has attracted attention. 3D imaging by confocal laser microscopes are widely used for observation of living cell. These microscopic techniques have a high resolution in the direction orthogonal to the optical axis (Z axis), however, there is a problem that the resolution in the Z axis direction is not high. In this research, we aim to construct a high resolution 3D image in all directions of XYZ with expanding the small micro hand system. This micro hand system could achieve high precision manipulation and control attitude of target object freely.

In this paper, we propose simple and effective microfluidic device to load a mechanical stimulus on numerous cells with scalability, in order to investigate the correlation between gene expression and morphological changes of a cell. Therefore, we make a microfluidic device which apply a mechanical stimulus (compression force) to cells by using MEMS technique. In particular, our improved device is designed to collect the numerous cells. In experiment, we achieved the compression of both cells and cellular nuclei, observing the deformation of each cells and cellular nuclei precisely, and collecting the numerous cells for performing gene expression assay. Furthermore, we designed S-curved microchannel and successfully find morphological changes of nucleated cell (mouse fibroblast cell) by repeated compression in one view filed by using high speed observation system differ from previous study.

In recent years, research targets in tissue engineering have been shifting from simple tissues such as muscle and skin to more complex tissues like living organs. Among living tissues, there are many types of tubular structures in vivo such as blood vessels, tracheas and intestine. These tubular structures have multiple-layered and wavelike surface in order to absorb nutrients or secrete hormones and other material efficiently, so we need to construct these complex tubular structures to reproduce their various functions. In this paper, we propose a novel device which integrates a pulsatile flow generating mechanism and a tubular structure generating device in order to construct complex tubular structures with wavelike surface like living tissues.

The average age of agricultural workers is aging. We focused on under-trellis work done on crops on the shelf with raising their arms. In the upward work, it is necessary to keep the arm with its own weight of approximately 2 to 4 kg. It is considered to be a cause of fatigue of arms and back. Therefore, in this research, we attempted to reduce the fatigue of the upward work by the weight compensation mechanism that balances the elastic force of the spring and their arms weight. It can compensate for their arms weight by the force of the spring in various postures of the arm at the time of work by using lightweight and highly accurate weight compensation mechanism authors developed. In this paper, in order to verify the practicality of TasKi at the agricultural site, we conducted a test survey simulating apple harvest under cooperation of San-nohe cho, Aomori Prefecture. As a result of the questionnaire survey, 80% of subjects showed favorable opinions on assisting method of TasKi.

This paper describes a method of recognizing “Ooba” (green perilla leaves) using deep neural network. We are developing a robot for ooba harvesting support robot. In order to manipulate ooba exactly, the robot needs to recognize ooba’s parameters such as width, height and direction using images. It is difficult to adjust parameters because the leaf colors change depending on the weather and the growth policy of each farmer. To reduce the time for adjusting parameters, we adopt deep neural network (DNN). We train DNN for recognizing ooba region in the image. We use image processing techniques to annotate labels efficiently. We also compare the performance of the method for sevral numbers of layers of the DNN.