A newly developed rollator integrated with 1'DOF rotating chest support is used to encourage an easily walking. In this paper, walking with the new rollator was measured using a motion capture system and an evaluation of improvement on the walking was made. The results show that body swing is easily extracted using the new rollator and a partial effectiveness on the walking assistance is confirmed.

View full abstract

In this study, a method to facilitate standing and walking by reducing sway in upright subjects was investigated. The technique involves the application of minimal fingertip force (also referred to as light touch contact, or LTC) with a virtual impedance partition surrounding the subject. In this set-up, a vibrotactile stimulator provides virtual force feedback when the subject comes into contact (known as virtual light touch contact, or VLTC) with the partition. To support the goal of creating a wearable aid that facilitates standing, the partition's impedance properties were examined in order to determine the extent to which sway could be reduced, and an unprecedented VLTC system with an acceleration sensor but no equipment for motion capture was tested. Based on this approach, sway was successfully moderated in healthy individuals.

View full abstract

This paper proposes a new class generation method for pattern classification problems. The approach enables estimation of a posteriori probability for learned and unlearned classes using a neural network, and allows evaluation of unlearned data distribution structures with a self-organization map (SOM) based on the probabilities estimated. By reconstructing the network structure dynamically, new classes can be generated from data of undefined classes via network training using given learning and evaluation samples. The method can be applied to various pattern discrimination problems such as electromyogram (EMG) classification. In the experiments reported here, classifiability with the proposed method was demonstrated using EMG patterns. The results showed that the technique provides a high level of performance for learned, unlearned and new class discrimination.

View full abstract

Recently, the proceeding of aging society has encouraged research and development of power assist systems. The authors' research group also studied on the control system for robust assist systems. It was mainly considered to be used in industries such as production factories. However, if such a cart is applied to welfare purposes, due to the uses of high specification computers and their batteries installed on the cart, the mass of cart becomes heavy to be used by aged people. Therefore, this study discusses a novel control system concept of the robust assist cart for welfare applications. Especially, as its first step, this paper discusses the downsizing and lightening of the cart on the basis of network control configuration using Wi-Fi and reveals the stability of the novel system in experiment and simulation.

View full abstract

Stair ascent is still difficult for transfemoral prosthesis users. So far, a passive knee joint unit for transfemoral prosthesis developed in our previous study. However, experimental results demonstrated that the users faced difficulty in the load response phase. In addition, large moment power of the prosthetic side's hip joint in the stance phase was required, leading to an imbalance of power consumption between the intact and prosthetic sides. Therefore, the purpose of the present study was to develop a passive transfemoral prosthesis for stair ascent, which had a knee extension function associated with a movable ankle joint during the stance phase in order to change the range of the force that can be exerted on the prosthesis by users to emerge the designed functions. Stair ascent experiments demonstrated that that stair ascending was achieved with the proposed prosthesis in the step-over-step manner. In addition, it was shown that the hip joint moment power reduced to equalize the joint moment powers in both sides.

View full abstract

The popularization level of the electric arm prosthesis in the present Japanese market is still very low due to the high cost and the lack of function, safety, etc. Our purpose is to know how to make cost-effective electric arm prosthesis to popularize it. In our previous research, we made an electric arm prosthesis and applied a balloon shaped soft pressure sensor's signal to control the opening and closing speed and grasping force of the finger. However we found that it was difficult to control open and close speed and grasp force of the finger well. We thought the sensor had two problems. First is that the sensor's output voltage gain is too high. Therefore, we required fine-tuning power to drive the electric arm prosthesis. Second is that the output voltage did not change from default position to pushing the sensor 2mm. Therefore the sensor's response was poor. Based on the above findings, we have investigated how to improve the linearity of the pressure sensors in the initial deformation area. After investigation, we found that the white color of the sidewall of the inner space of the balloon made a diffuse reflection. We made a refined sensors that has black inner sidewall and white inner ceiling at the top side. By using this refined sensor and decreasing the gain of the amplifier, the response in the initial deformation area was improved drastically and we got the much better controllability of the finger opening speed and finger grasping force. Furthermore, we made a wrist-unit to add rotational function of hand and drove it by the additional pressure sensors to measure the movement of the ulna and radius. The operator can control well both grasping function and twisting function simultaneously.

View full abstract

This paper is concerned with an advanced transfer trajectory control of overhead traveling crane. In the overhead traveling crane, it is required to reach the target position in a short time, avoid the obstacles, and suppress the load sway. In this study, it is verified that a practical motion of the overhead traveling crane can be represented with the linear model. And, the transfer trajectory of the cart in the overhead traveling crane is designed in consideration of sway suppression, obstacle avoidance, and rapid transfer by the trajectory planning which has been proposed by anthers in recent study. The effectiveness of the proposed trajectory planning is verified using the overhead traveling crane.

View full abstract

The extensibility and flexibility of the industrial robot system can be improved by installing external PC which generates the trajectory based on external sensor such as force sensor. We have already built a "High-precision finishing system" which has an external controller. By using external PC, the force control can be realized. Thus the system can supersede manual labor which cannot be done by conventional robot system. To enhance the performance; speed and accuracy, it is necessary to design the controller considering system dynamics. In this paper, we identify system model of the "High-precision finishing system" and verify the some major control method using identified feature.

View full abstract

In order to improve the vibration suppression of object swing, 2-degrees of freedom control system is constructed in this research. It is unified with the obstacle recognition system and path planning system which are proposed until now, and the usefulness of the autonomous overhead traveling crane system integrated was confirmed.

View full abstract

This paper presents Petri net based models of autonomous mobile robot tasks. Motion tasks are modeled by Petri net modules and organized in layers with different degrees of abstraction as basic building blocks, or macro places, which allow a hierarchical view replacing a block by its corresponding net model. Hierarchical net decomposition provides a set of small net models composed of autonomous activities by several active components or effectors in the system, which are represented by transitions. Besides navigation control, cooperation for special subtasks requiring the exchange of synchronization signals is considered. In the models, states set by communicated messages are seen as resources and represented by Petri net places. Thus, cooperative tasks are modeled by adding extra information to the individual task models to account for synchronization by explicit communication.

View full abstract

This paper clarifies the dynamic behaviors of a mass measurement system with conveyor belt (a checkweigher). The checkweigher can be approximated by a spring-mass-damper system as the physical model, and the equation of motion is derived. The validity of the proposed model can be confirmed by comparison of the simulation results with the realistic responses. Moreover, the transient responses of checkweigher with Analog and Digital PID controller are compared.

View full abstract

Bilateral control is a teleoperation system enables to transmit the information about force sense. And, hydraulic actuators of high power weight ratio are often used in ultimate environments. Open loop hydraulic circuits, which are generally used, contains servo valves in the structure. But using servo valves are the cause of the low backdrivability. However, the robot can have high backdrivability using close loop hydraulic circuits with servo pumps. In this research, we use an electric actuator as a master robot and a hydraulic actuator made up of a close loop hydraulic circuit as a slave robot and bilateral control is realized.

View full abstract

This paper presents a 10-ton hydraulic press machine installed with a hydraulic hybrid servo booster proposed by the authors. The experimental results show that the press machine can move the slider at upto 650 mm/s and regulate the position and force errors in 1 micrometer and 400 N, respectively, recording competitive scores to existing commercial servo presses.

View full abstract

This paper reports on a hydraulic quadruped walking robot, which is designed for the prototype of leg wheel type robot RL-R1. The legs are operated by hydraulic cylinders and rotary actuators. The range of motion and joint torque are set aiming at forest operations or disaster response. We achieved lightweight and high strength leg by using carbon fiber reinforced plastic. This paper shows some preliminary experimental results including squat motion under 165 kg payloads and swing motion at maximum speed. As a result, the developed legs are confirmed to have sufficient torque and strength for our mission.

View full abstract

Mechanical design for legged robots is extremely challenging due to the trade-off between the strength and the weight. To provide the testing method for those robots, we propose to use hydraulic excavator as a loading machine that can handle a wide range of configuration. We show that the excavator can control the load force to the robot by using pressure feedback based torque control, not only can stably support the robot swinging its leg very fast.

View full abstract

We propose a power-assist system using constrained predictive functional control considering set-point trajectory. The system is capable of realizing the motion-assist safely while predicting operation touque and future trajectory on-line under output constraint Some experimental results of verification are given in order to verify the usefulness of proposed power-assist system.

View full abstract

The aim of this study is to develop a motion control strategy of a walking support robot based on the estimation of user's state to stabilize the posture of the user in a walking state. In this study, the motion control strategy based on the detection of the user's legs using a laser range sensor (URG) was constructed to improve the quality of the walking rehabilitation. To analyze the walking state of the user using both the robot controlled by the proposed motion control strategy and the rollator which is a general rehabilitation equipment, the walking experiments were conducted. The experiment results showed that the postural stability of the user in the walking rehabilitation was improved.

View full abstract

We developed a mobile power assist equipment as a human-coexistence robot. We discussed how to develop the power-assist arm and moving mechanism for power-assist equipment. Additionally, we discussed how to reduce the controlling torque of the power-assist arm mechanism through mini-max optimization and manufactured a mock-up of the power-assist equipment. In this paper, we report on how to control the mobile power-assist robot in considerable situation like as the control target, the circumstance, disturbances and restrictions of the control law.

View full abstract

In our research, we aimed to construct a strategy to plan a path of the robot that includes the inducement of humans depending on environmental factors such as road width, position of obstacle, position and posture of humans. By using RRT (rapidly exploring random tree algorithm), a probabilistic path-planning calculation method, we proposed constructing moving strategy based on the environmental factors. We implemented the proposed strategy into the robot and conducted experiments in several environments. From the experiment, we saw the robot move and induce the human accordingly to the environment, and the usefulness of the proposed method was demonstrated.

View full abstract

In this paper, the results from the experiment show that human inducement done by changing the values of parameters of the arm movement of the robot changes the evaluation of human movement. Analyzed results from the experiment are, the longer the distance becomes for when the robots starts its signal to the human, the avoiding of the robot by the human starts from a greater distance, when the speed of the arm movement is constant the efficiency loss becomes smallest, and when the movement angle becomes larger the avoidance distance becomes larger and also the loss of the moving efficiency becomes larger.

View full abstract

Robot path planning based on inducement of humans by touch requires a model of human reaction against robot's inducement. In this paper, we aimed to make a human trajectory tendency model for inducement by contact to be implemented into collaborative movement path algorithm. In order to achieve this, we conducted 2 experiments. First, we placed a human in the place of the robot to observe human reaction against inducement by touch to determine parameters that affect human movement. From the parameters taken from the first experiment, we then conducted the second experiment with the actual robot. From the results, the direction the robot touches the human and touching by the robot's arm surface has an effect on the human to move in the desired direction.

View full abstract

The purpose of this study is to investigate the activation state of the higher brain function about the positioning task added friction properties in the human-robot cooperative task. Experimental conditions are classified into the following three kinds of tasks; (Case1) Positioning task without friction properties, (Case2) Positioning task with friction properties, and (Case3) Positioning task added friction properties with larger frictional force than Case2. The experimental results showed that there were significant differences on the activation of a motor area and a somatosensory area and that there were no significant differences in the higher brain activity level among the three kinds of tasks.

View full abstract

Hand gestures can express various kinds of information; however, exiting techniques have problems for practical use such as environmental limitations and obstructions to natural gesture. To address these problems we have developed a wrist contour measurement device that recognizes hand shape and pronation angle in the previous work. The previous paper indicates that it is necessary to increase the number of recognizable hand motions to realize a high-usability screen operation. This paper reports the trial of simultaneous recognition of hand shape and wrist flexion state. We observe wrist contour and make clear the phenomenon, design features, and construct a recognizer. The simultaneous recognition rate of four hand shapes and nine wrist flexion states achieved in this research is 75.5%.

View full abstract

We have been developing a robot which work in cooperation with workers in the automobile bumper assembly line, named B-PaDY. This robot is not directly involved in the assembly task but does non-essential task for workers. In this paper, we propose a method to decide the support timing based on logistic regression model and confirm its effectiveness by performing experiments. By recognizing the worker's behavior and supporting the worker at the proper timing, this system improves work efficiency and supplies the bumper to the worker effectively. The proposed model is used for modeling of robot's decision. It is made by machine learning that uses the worker's behavior as input.

View full abstract

The purpose of this study is to propose motion control algorithm of a walking support robot which is controlled in accordance with actual walking rehabilitation data obtained from the motion of the walking rehabilitation using parallel bars. To analyze the walking state of the user using the parallel bars which is a general rehabilitation equipment, the measurement experiment was carried out. The proposed motion control strategy is developed to simulate the motion of the walking rehabilitation. To evaluate the proposed strategy, the walking data using the walking support robot was evaluated about the results of the moving fluctuations of waists, the reducing rate of the weight and the rate of the zero moment point (ZMP) in the base area of the support. The proposed algorithm brought the result which is same effect in the case of the walking rehabilitation with the actual parallel bars.

View full abstract

In this paper, we present a mechanical design of an exoskeleton-type power assist suit that helps a wearer to lift up and transport heavy objects while walking. The assist suit supports heavy objects by itself thus the force exerting on the whole body of the wearer is reduced. In addition, the hip and leg joint axes are aligned with those of the wearer so that the wearer can walk smoothly. The experimental results show that the developed power assist suit is able to support and transport heavy objects while walking without exerting a load on the upper body of a wearer including the arms and legs.

View full abstract

While seamless switching between the accompanying and handling modes for an outdoor personal robot is required, a lack of sensors that can robustly detect the proximity and touch of the user has hindered its practical development. In this study, a novel sensing technology using electromagnetic field fluctuation is presented. The experiment, using a suitcase with the sensor, showed that the area surrounding the user, as well as the direct touch of the user can be separately detected. We also embedded the system into the robotic suitcase and showed that the robot can robustly follow the users.

View full abstract

In recent years, robots have become familiar to human society. Human beings are doing the cooperation to act in anticipation of opponent behavior and intention in daily life. But, household robot that is currently popular is mostly what it is not possible. Humans have gained a lot of information by the visual information. Thus, by analyzing the human gaze movement, it was possible to estimate the human intention and the state in relation. If this reflected in the behavior of the robot, the robot would run the appropriate support action for human. In this paper, we propose a method for measuring the gaze point of multiple persons. As a result of propose a method is possible to perform the measurement point of gaze for multiple persons.

View full abstract

Machine operation assist could be improved by visual and auditory information given the time difference in presentation. In our study, the validity of multimodal assists with time difference is experimented by vibration control task of a pendulum. 4 types of operation assisting was compared time to complete vibration control. The operation assists are (i)without assist, (ii)with only visual assist, (iii)with visual and auditory assist (no time difference) and (iv)with visual and auditory assist (100[ms] delay for the auditory). Experimental results show that giving a time difference to the assist presentation is effective for the machine operation.

View full abstract

This paper addresses a skill assist system by change software, focused on the initial stage of proficiency, the human used visual cortex. The purpose of the basic task skill assist with visual information. Most of the conventional assisted systems have been required specialized machine and low prevailing versatility. For the proposed method in the first stage are experimental tasks verify of human characteristics by visual information are investigated. These experiments give the visual information for assist to the bar tracking task. Finally, the present study has demonstrated that operation characteristics are changed by visual information, The results suggest that the visual information is valid for skill assist.

View full abstract

Unlike industrial robots, human interaction robots are designed to focus on coexistence with humans. In order to coexist with humans, there are many considerable factors. Among them, one of the important factor is that how the robot is shown to human. If the presence of the robot is too weak or too strong for the situations, the human may feel uncomfortable. In this paper we propose Multi-legged Giant Yardwalker (MGY) which can vary its presence by changing its height. The MGY is able to change its presence properly based on the situation, and the user likely to gain consciousness of the robot easier, or may not be aware of the robot. The early experimental results showed that the robot expressed various presence by height difference.

View full abstract

Recently, simple instruments to measure the brain waves (EEG) is being developed and commercially available. In addition, a radio controlled helicopter controlled by the measured brain wave data is sold. The final goal of this study is to develop swarm robots controlled by EEG. Measuring instruments for medical use are very expensive and restrict human motion. Contrary, simple and easy to ware devices are commercially available that is inexpensive. Therefore, we choose a simple brain wave measuring instrument called B3band, which is able to send the brain wave data to PC through Bluetooth. In this paper, we describe a concept of our control system using B3band, and examples of recorded data. Also we introduce a mobile robot which has two driving wheels with RC servos.

View full abstract

The purpose of this study is to investigate the effect of interaction between a leader and a follower on the higher brain function from the relation between the force applied by the operators and the activity level of the higher brain function. Subjects are labeled by the following three kinds of method; [Task 1] Subjects aren't designated the leader nor the follower, [Task 2] Each subject are designated the leader or the follower, and target points are shown to both of them, [Task 3] Each subject are designated the leader or the follower, and target points aren't shown to the leader but not to the follower. Higher brain function activity level in motor area is measured when operators perform above mentioned the three tasks. The experimental results showed that there were significant differences in the higher brain activity level between the leader and the follower.

View full abstract

This paper presents a development of a mechanical device to adjust the dynamic characteristic between a human operator and a robot under their cooperative task. The proposed device has a mechanism which can provide an adequate dynamical interaction between the robot and the human operator to stabilize the robot control system in a constraint state. To develop the control strategy of the mechanical device, the relationship between a controlled variable and brake force is revealed in the brake force calibration experiment. To confirm the effectiveness of the proposed mechanism, a human-robot cooperative experiment is carried out. Experimental results show that the proposed control scheme is effective for stabilizing the human-robot cooperative task system.

View full abstract

We are aiming to automatic generation of high human behavior model reproducible from huge data about human behavior. There-fore, it is necessary to extract a part of the basic elements of a series of time-series data. In this paper, a series of time-series data is quantized in the HMMs, and to network the transition between high reproducibility HMMs. Extracts the basic elements based on the network to generate the behavior model.

View full abstract

The current, is progressing aging was built bridges and tunnels more than 40 years ago. but Maintenance engineer is also missing. Inspection of the runnel requires traffic regulation. Maintenance engineer carried out as close as possible to the inspection point by mobile elevating work platform. But, take the time to visual inspection and hammering test for inspection area is wide. And, there are variations in the inspection of visual and hammering by human. We proposed a "variable guide frame vehicle for inspection of tunnel lining concrete" for new maintenance technology of infrastructure. We report that inspection plan and basic constitution of the system in this paper.

View full abstract

To increase the speed of automatic transporting vehicles in underground narrow roads, we propose a differential four-wheel vehicle that allows contacts to the wall using roller bumpers. A one-tenth scale model is developed for a feasibility study. Using the model, basic experiments to evaluate the impact of wall-contact have been conducted. Wall-contacting running is realized by offsetting the target path to wall. The results of the experiment reveal the possibility of the travel utilizing the wall without a large impact. The additional basic experiments have been conducted to evaluate the running in situations where position noise is added. The results of the experiment showed that driving utilizing the wall as guide structure enables a stable travel even in situations where position information contains severe error.

View full abstract

It is quite dangerous for human workers to work inside Fukushima Daiichi Nuclear Power Plant where many radioactive obstacles and debris exist. There, various robots are deployed to address many tasks, but it is not easy for even robots to realize smooth locomotion and efficient operation because radioactive debris and architectural structures such as steps and doors interrupt them. In addition, utilizable sensors and actuators are restricted in such an environment. To overcome those difficulties, we have proposed an automated construction system of robot locomotion and operation whose rail structure is modularized. The structure is anchored to walls or floors and enables working robots to reach a specific destination. This paper describes development of rail structure supporting mechanism. The feature of the developed mechanism is that it uses resin foam for three different functions; actuator, adhesive material and structural material. Basic experiments are conducted using the developed prototype. Experimental results verify the feasibility of pillar expansion, anchoring and fixation motions, and they also reveal the impact of components' properties on the supporting performance.

View full abstract

Ensuring safety of the Operator of the Construction Machinery is an important issue in the disaster restoration spot and Danger environment. Construction Machinery of the Remote Control is used in the environment where a human being cannot enter. It is thought that the Remote Control greatly has decreased Operating Efficiency in comparison with boarding operation. A lot of causes of the drop of the Operating Efficiency are thought about, and one of the big factors is transmission delay of the images. Therefore the authors built a Remote Control system of the Hydraulic Excavator using the spot radio and public cable broadcasting network to confirm a problem of the Remote Control. And a system used Remote Control of the built Hydraulic Excavator and performed inspection about the transmission delay time for picture. This report introduces delay time of the video transmission with a built Remote Control system.

View full abstract

The influence of the error on throwing motion will be reduced by an appropriate trajectory design. In this paper, we focus on the human-error in zero-adjustment of joint angles of a robot. Based on the conventional approach, the calculation method of error sensitivity in planer throwing and the robust trajectory design are explained. A 3-DOF throwing robot is prototyped, and human-error of zero adjustment is measured and analyzed. An error ellipsoid is introduced based on the measured data, which is utilized for calculation of sensitivity. By Monte-Carlo simulations, the effectiveness of considering error distribution is evaluated.

View full abstract

In automobile assembly process, there are still a lot of processes which need to be controlled and carried out by human beings. In these processes, we considered how to be more effective in performing the assistance to workers by partner robot rather than to complete the whole work by only robots. In this paper, we propose a partner robot that operates in the same workspace as a co-worker in the bumper assembly line. The Robot co-worker for bumper assembly line named B-PaDY hands over a bumper to a worker at the proper timing. B-PaDY is an approach to improve the work efficiency by carrying bumper to a worker. In particular, prototype design and its concepts are described in detail in this paper.

View full abstract

The purpose of this study is to investigate moving and pointing behavior characteristics during human-robot cooperative positioning tasks. In previous studies, various variable impedance control systems, in which impedance parameters are adjusted following a projected trajectory, have been proposed. However, only characteristics of trajectory during moving behaviors were evaluated, without mentioning the relation between moving and pointing behaviors. In addition, the difference of task tune and a distance in each experiment is not considered for the simplification. Therefore, in this study, the relation between pointing accuracy and a trajectory of velocity is analyzed using a correlation efficient and velocity profiles in various conditions to reveal the task behavior of the human operator in the human-robot cooperative task system.

View full abstract

In recent year, force feedback technologies which feedback force information to operator are developed. However, in this study, we use of force feedback for directly assist. Therefore, force feedback joysticks are developed and proposed force feedback assist method is verified. The neural network learning is utilized for the force feedback joystick. Force feedback joystick is constructed by DC motors and motor drivers and a microcomputer. Experiment result is to assist force feedback at lateral is valid and twist is not valid. Therefore experiments of force feedback assist for neural network learning is conducted at lateral. Finally, we experiment about evaluating of neural network learning accuracy with force feedback assist.

View full abstract

Recently industrial robot systems are introduced widely even in the small and medium size enterprises. However, it is difficult for them to employ professional engineers permanently. Then, they need to depend on outside to maintain industrial robots. Therefore a remote operation support system for industrial robot must be useful to do maintenance and solve problems immediately. In the previous research, a remote operation support system by means of cognitive audio-visual infocommunication has been provided. In this system, it was considered to be a problem to hear plural audio information simultaneously. This paper investigates how an information presentation using vibration stimuli is useful for operating an industrial robot. Especially, we compare users' performances and mental workload between on-site operations and remote operations. The experimental results show that the information presentation can provide information, also reduce a workload more than when using only visual information for remote operations.

View full abstract

This paper proposes a visual inspection methodology for the ceiling element of the gymnasium utilizing the blimp robot with the Wi-Fi camera. In the proposed methodology, the blimp robot with the Wi-Fi camera inspects the damage location and captures the photographic image of the damage condition. The inspection blimp robots will be able to estimate the damage condition without any process of engineers' on-site-inspection involved. To demonstrate the capabilities of the blimp robot with the Wi-Fi camera, the blimp robot is applied to estimate the ceiling of a gymnasium. The results of the application to real ceiling indicate that the Wi-Fi camera on the blimp robot can capture and estimate the image of the condition of the aging ceiling.

View full abstract

Nowadays, social infrastructures in Japan are rapidly becoming old and many facilities which help Japanese people to make a good life have various problems. We have to immediately develop new technologies which enable us to cope with those problems efficiently and effectively. The objective of this research is to develop two key technologies. One of them is a means which enable us to take clear movies in turbid water, and the other is an attitude control method of underwater robots because the robots have to go to designated points. First, we designed Anchor Diver 4.1 for developing die attitude control. Second, we made a new underwater robot named "Anchor Diver 5.0" which has a clear case and a water proved camera. Then, we show results of practical experiment at Miyagase - dam in Kanagawa prefecture and evaluate the means we adopted in this research.

View full abstract

In order to realize automatic visual inspective systems for infrastructure facilities, this paper presents an automatic path planning method for UAV(Unmanned Airial Vehicle) based on 3D structure models. 3D model described by StL file format has normal vectors of faces. By using the information of normal vectors and positions of the faces, the inspection points are calculated. An inspection path including all inspection points is generated by searching algorithm. The proposed method is evaluated on computer simulation.

View full abstract

In Japan, many infrastructure has been aging. It is necessary to maintenance and inspection of them, but it costs too much human resources and money. This research aims to the robots replace human in bridge inspection tasks and removing sediment from shoe of bridge, we developed an 8 cable-suspended robot for bridge inspection. In this paper, we describe the robot design and the results from preliminary mobility test.

View full abstract

Aged deterioration of social infrastructures such as concrete tunnels has become a serious problem. For safety and security, reliable inspection of them is highly demanded. Hammering test is one of efficient inspection methods. However, the inspection task needs a lot of expert workers. In this paper, a methodology of constructing a hammering-inspection-based defect detector that can be applicable to concrete structures is proposed. Furthermore, the diagnosis process based on a new boosting algorithm is visualized for the purpose of supporting reliable inspection. The proposed defect detector was applied to signals obtained in a real inspection site of a concrete tunnel.

View full abstract

The problem of existing teleoperated construction machineries is work efficiency compared with directly operated ones. In fact, principal factor of inefficiency of the systems have not yet clearly determined. But normal 2D images cant give a sense of distance enough. So, other researchers proposed to use operator's eye with 3D vision or the other angles view. However, an operator tiring to be using the 3D vision. Even if there are multi displays, the operator hardly only use one display. So, this paper proposed to create 3D environmental map and to superimpose the ground shape on the operator's eye images. The superimposed image assist the sense of distance from the work point to the bucket tip on the 2D image. Depth image from the sensor on the miniature excavator was projected to the global coordinates and eliminated the conflicted point with the body. The 3D environmental map was estimated with bayesian inference and saved to the Digital Terrain Model. Drawing a uniformly spaced grid lines to the terrain surface and projected the bucket tip to it. These processes were processed on CUDA and GLSL. The system was evaluated on the miniature platform.

View full abstract