In unmanned construction for landslide recovery, it is important to know the strength of the ground to prevent construction machinery from getting stuck due to the softness of the ground. However, a problem with wide-range ground strength estimation in previous studies is that unevenness and inclination variation of the surface can reduce the accuracy of the estimation, so this study aims to propose a cone index estimation method that is robust to unevenness and inclination variation. The proposed method uses a reflectance model to estimate spectral reflectance from spectral images considering the poses of the light source and the camera with respect to the ground surface, and then estimates the cone index from this spectral reflectance. The results of indoor validation experiments confirm the effectiveness of the proposed method.

We are developing a robotic system that suspends a spherical helicopter by multiple wires for safe and highly efficient inspection of glue surfaces and bridges. This system takes the form of a UAV suspended by tethers from different directions. Minimizing tether tension is an important issue because tether tension affecting a spherical helicopter, especially when operating in "flight mode," causes flight envelope limitations and payload limitations. Depending on the installation environment, it is possible to suspend a spherical helicopter with more than two tethers. If the number of tethers suspended from different directions is increased, it is expected that the tether tensions on the spherical helicopter will be canceled out and the tether tensions will be kept small. The purpose of this paper is to propose and verify the principle of a method to minimize the effect of three or more tethers on the spherical helicopter by canceling each other's tether tension.

This study aims to develop a unified detection method focusing on ”anomalies” caused by attacks. Security is one of the most concerning issues as robots become more prevalent in society. Recent robots exchange various data via networks and operate in the same space as humans. Therefore, the possibility of damage caused by malicious attacks is increasing, and robot security measures are becoming more critical. However, it is difficult to comprehensively prevent all attack vectors, because robots need to deal with a wide range of attacks from both physical and cyber spaces. In this paper, we propose a unified method for detecting anomalies using a virtual environment that can observe the actual environment. Specifically, anomalies are detected by using the differences between observation information, and the type of anomaly can be identified by analyzing the time series changes in the differences.

We have been developing a welding robot aimed at repairing half-collapsed buildings in disaster area. In this study, we propose a welding system that performs fillet welding on a curved surface with a 6-DOF robot arm by remote control. When a welding path is roughly specified on the image captured by the RGB-D camera attached to the robot, it is possible to correct the welding path using a touch sensor. In addition, the system is developed using ROS for efficient development. In the experiment, we confirmed that the developed system is able to perform fillet welding on a buckled plate remotely.

An efficient method to determine the bolt fastening condition of bridges is required to replace the visual judgment of workers. Therefore, this study proposes a wearable discrimination system using images. In this paper, we examine the effects of the viewpoint and shooting distance during the work on the discrimination results.

The constructions in earthwork environments, such as lunar and disaster sites, often involve extensive and hazardous work with heavy equipment. In particular, to prevent secondary disasters and to adapt to changes in the environment, the multiple robotic operations are expected to be highly effective. Several systems have been developed to address this, while they still have not reached to real systems due to the complicated field properties and the issues of how multiple-heterogeneous robots effectively cooperate. In this study, we propose a design concept for a system architecture that can be efficiently researched and developed to use multiple heterogeneous construction robots in earthwork environments. Realistic simulation experiments with both earthwork properties and robotic characteristics have shown the applicability of the proposed system architecture.

In recent years, construction demand has increased although the number of construction workers has decreased. To solve this problem, The initiative of Digital Transformation (DX) has been active, and research and development on automation of hydraulic excavators has been carried out. It is difficult to set the operation in advance to excavate a stable amount of soil safely because the quality and shape of soil vary. It is expected that excavation paths can be generated according to the quality and shape of the soil by adapting reinforcement learning to the generation of excavation paths for autonomous excavators. This paper describes the results of applying reinforcement learning to the generation of excavation paths for autonomous excavator.

The civil engineering and construction industry has seen an increase in occupational accidents in recent years. Although several systems to improve the efficiency of earthwork operations have been developed, they have not directly involved in improving safety. The objective of this study is to explore the use of a cyber-physical system (CPS) to improve both the efficiency and safety of earthwork operations. The CPS platform should have a function to store and manage environmental information in a database. The ROS2-TMS, which is the CPS platform for service robots, satisfies this requirement. However, the ROS2-TMS cannot handle environmental information specific to earthwork sites, such as terrain and ground data. Therefore, in this research, we developed an improved version of ROS2-TMS named ROS2-TMS for Construction that can store and manage site-specific environmental information in a database. Additionally, we developed a VR interface to visualize the stored environmental information.

In mountain tunneling, the subject of my research, dynamite, and excavators are used for excavation. After excavation, concrete is sprayed to prevent the tunnel from collapsing. In the spraying operation, concrete must be sprayed to a specified thickness, and the operator visually checks the thickness of the concrete while operating the concrete spraying machine. There are concerns about dust damage to the workers and falling, so unmanned operation is required. Therefore, our research group is developing a technology to control the thickness of concrete spraying. The method of the technology we are developing requires a positional posture of the sprayer nozzle, but there is a problem that sensor equipment, etc. is easily broken due to the adhesion of concrete dust. I will estimate the position of the sprayer nozzle using a camera and INS, which can simplify the failure response, and report the experimental results.

The purpose of this experiment was to investigate the effect of the swing operation difficulty on operator stress. Four swing operation difficulty conditions were set, using the swing operation difficulty model. During the experiment, sense of agency, and stress were measured. As a stress index, we used the LF/HF ratio obtained by frequency analysis of the electrocardiogram. We investigated the relationship between stress and the swing operation difficulty. As the swing operation difficulty increased, the sense of agency decreased and the LF/HF ratio remained unchanged. In future research, we will review the experimental conditions and task settings so that we can observe the stress tendencies of the experimental participants.

Continuous tracking of the tip position is necessary for uniform compaction of the concrete vibrator, but a simple measurement cannot pinpoint the exact position. In this study, we analyzed the relational expression between the tip of the vibrator and the position of the hand, measured the x-y coordinates of his hand using the RTK-GPS positioning value and the laser distance sensor, and used the inclinometer to measure the z-coordinate. was measured. In addition, by displaying the area affected by compaction in real time using the measured coordinates of the tip position of the vibrator, we were able to estimate the compaction area of concrete by the vibrator, contributing to the reliability of construction.

The authors are developing a system for detecting anomalies associated with the deterioration of piping equipment in a geothermal power plant. In this study, we proposed a method of constructing a three-dimensional point cloud model of outdoor infrastructure by combining SfM(Structure from Motion) point cloud as a method of complementing the parts that cannot be observed by LiDAR(Light Detection And Ranging) point cloud. By using marker features, it is possible to generate point clouds with an error of approximately 31 mm without using GCP(Ground Control Point) that require surveying. This makes it possible to construct a highly accurate and high-resolution three-dimensional point cloud model of outdoor infrastructure from LiDAR and SfM point clouds.

In construction industry, to operate efficiently construction machineries, three-dimensional obtainment of environment of the sites is quite important. In particular, position and pose of the construction machineries around the self-machinery is necessary to achieve tasks toward other ones, such as soil-loading work to dump truck with backhoe. Recently, digitization of the construction sites with ICT technologies is promoted, and remote operation is desired. It is difficult, however, for remote control operators to comprehend the situation accurately. In this study, we propose a system to assist the operators by providing three-dimensional situation in the construction sites. The proposed system measures the environment visually and three-dimensionally and detect machineries at a site, and their position and pose are estimated. In an experiment with actual construction machineries, evaluation of performance of the proposed system was conducted.

In the construction industry, recent developments in ICT technology are expected to lead to the digitization of sites, such as remote operation of construction machineries. In particular, excavation work by hydraulic excavators is indispensable at construction sites. To achieve this work efficiently, it is necessary to comprehend environment of the sites three-dimensionally. Furthermore the terrain needs to be obtained without any blind spots. Therefore, we propose a method to measure the terrain around a machinery with multiple LiDARs mounted on itself. The proposed system complements blind spots by integrating point cloud data obtained from the LiDARs mounted on the cabin and arm. The proposed system was validated by an experiment with actual machineries in a construction sites.

In recent years, robot technologies, especially UAVs, are actively introduced at the civil engineering field. One of the typical applications of UAV is the damage detection of the bridges etc. However, there are few examples of the indoor damage detection of the building using the robot technology. This paper presents the indoor crack detection methodology utilizing UGV and YOLO. In the proposed methodology, UGV captures many pictures or movies of indoor wall for detecting the cracks on the wall. After taking images of the wall, cracks on the wall are automatically identified by using YOLO. For the navigation of the UGV, LiDAR and AR markers are employed without SLAM. The usefulness of the proposed method is confirmed by experiments targeting university buildings.

The pico-EV・Eco Challenge is a competition in which the vehicles which carried a person run during 20 minutes using a low powered battery (7.2 V, approximately 1000mAh).This competition was held for the first time in 2012. The 2020 competition was canceled and the 2021 and 2022 competitions were held online. The 2023 competition was held as a hybrid of on-site and online. In this paper, trends and problems on this competition are clarified. And, , the motor, the reduction ratio, the body specifications, and so on about the vehicles on the pico-EV Eco Challenge 2022 and 2023 were introduced.

In recent years, along with increase of seafood consumption, land-based aquaculture draws attention for its low environmental load and less location restrictions. In fish aquaculture plants, measurement of fish size is important for growth management. In current system for land-based aquaculture, however, the measurement for which the fishes are landed is heavy load for both the fishes and plant workers. Thus, underwater body measurement is highly demanded. However, the underwater measurement is quite difficult task because rearing water of fish tanks becomes turbid easily. In particular, to measure fishes in the turbid water, a method to capture fishes from near distance with wide field of view is necessary. In this paper, a system to measure fish bodies in the turbid water using multiple stereo cameras is proposed. In an experiment in an actual land-based aquaculture plant, the performance of the proposed system is validated.

This paper presents a mathematical model of a robot with fin in an underwater environments. The model takes into account the hydrodynamic forces acting on the fin and the robot’s body. The robot with single fin and actuator is used for the experiments. The joint angle and the force acting on the actuator are measured to compare between simulation and experiment.

In this study, a submersible geared servomotor was developed to realize a penguin-mimetic underwater flapping-wing robot. The mechanism was composed of a commercial brushless servomotor and a lab-made plastic planetary reducer. The required rotational speed and torque were determined based on the previous data of actual wing kinematics of swimming penguins and hydrodynamic calculation of the lift and drag for the wings and body. The fabricated mechanism was 66 mm in diameter and 53 mm in length, weighing 0.35 kg. Underwater experiments demonstrated that the output power of the mechanism was 107.4 W with 160-rpm rotational speed and 6.41-Nm torque. The positional accuracy was as large as 2.7 degrees, with simple proportional control only.

We propose a marine animal tracking method for an autonomous underwater vehicle that utilizes multibeam imaging sonar. To estimate the position and pose of the target, we employ a tilt control device and neural networks. Our approach is evaluated on a turtle replica in a tank, and we demonstrate the efficacy of our method using simulations by integrating a particle filter algorithm for improved tracking.

Environmental recognition for underwater environments remains an unsolved technical problem. Currently, one of the critical issues in an underwater situation is the deterioration of the hull. In this study, we aim to detect the deterioration of the ship bottom automatically based on semantic segmentation on images captured from an acoustic camera.

Robot programming education has been actively promoted around the world. Robot programming education is recommended to incorporate collaborative learning, and the role of the teacher is important. However, due to educational disparities, teachers are currently unable to adequately support collaborative learning. We developed a voice support system that automates the teacher’s voice and a matching system that automates the group formation in online pair collaborative learning of robot programming, followed by an evaluation experiment of these systems. The voice support system had a positive impact on learners to solve problems. In addition, the matching system was highly satisfactory for learners with low basic academic ability.

Since the global outbreak of the novel coronavirus infection, many people have been forced to live under self-restraint. As of March 21, 2023, a policy to relax mask-wearing has been announced, and the situation is gradually returning to pre-pandemic life. However, due to the emergence of variant strains with high mortality rates among infants and elementary school students, children in local communities are refraining from long-distance travel and leisure activities. This study aims to contribute to creating an environment where children in local communities can live with hope and foster interest in robot-mechanical technology among preschool and elementary school children by developing and operating a mini railway vehicle with a 127mm gauge. Specifically, we held mini railway operation events and surveyed participants' awareness through questionnaires to examine the potential contribution of the mini railway to robot-mechanical education, and report the results of our investigation.

We are teaching "Fundamental Mechatronics Experiments," which is a required course for second-year students at this university. For the 2021-2022 academic year, we implemented a system where students were required to watch 5-15 minute lecture videos explaining the experimental principles, in addition to reading the experiment guidebook, before the actual experiment. As a result, while some students watched the videos and gained a sufficient understanding, there were others who did not watch the videos or watched them without gaining a sufficient understanding before the experiment. Therefore, improving the video viewing rate became a challenge. Starting in 2021, we created videos explaining the experimental principles using the "slow explanation" technique commonly used on YouTube and allowed students to watch them at their own discretion, conducting surveys to investigate their usefulness. In this report, we will present the results of the survey conducted in the 2022 academic year. We will also report on our findings regarding the use of short videos, which have been gaining increasing viewership in recent years.

It is not always possible for students who have learned machine learning in lectures to apply machine learning techniques to solve problems in the real field. Problems in the real field have various external factors and cannot be solved as easily as the machine learning practice problems dealt with in class. To fill this gap, this paper proposes an educational program that applies machine learning to problems close to real field problems. This educational program applies machine learning methods to the problem of autonomous navigation of a ship robot in a real sea area with many disturbances. For students to learn efficiently, the development platform is unified with MATLAB in this educational program.

We have been developing a robot with an intuitively understandable instruction set using a microcomputer, and have used it for introductory programming education at universities. In the classes, students not only created programs individually, but also collaborated to make the robots move simultaneously and cooperate with each other. In the collaborative work, we have continued to set and improve tasks in order to increase learners' interest and understanding of programming. In this paper, we introduce the various cooperative programming education methods we have used so far, and describe their characteristics and improvements.

This paper describes a small and inexpensive controller using Linux for control experiments and practical training of robots. Inexpensive components were selected for ease of use in practical training. The power supply circuit is designed to be used with servo motors. A base board was made to integrate necessary functions.

In the future, the size of the domestic robotics market is expected to grow significantly, especially in the service and manufacturing industries, and the field of robot engineers is expected to expand. Our university is committed to fostering engineers and researchers in the fields of mechatronics and robotics who can respond to these needs. We have introduced mechatronics education in the form of competitions using 2-axis robots in the third-year seminar course, where students can learn electronic circuit design, control programming, and mechanical design through robot development. This paper describes the contents and issues of the practice course, which is in its fourth year of implementation.

While AI technologies are being developed in a variety of fields, there is concern about the shortage of AI human resources to carry them out. In learning about AI, it is important to understand not only the algorithms of AI, but also how to utilize AI. In university education, there are few opportunities to learn practical application of AI to other fields. To develop AI human resources, we created a class combining AI and robotics for first-year undergraduate students in engineering. In the class, students create an object detection AI model and control the drone's movement by detecting markers with the model.

In a scientific laboratory, students can get more technical knowledge and skills since they had taken lectures, experiments and practices in a classroom. In this paper, it is suggested that an evaluating method of engineer abilities for students in a laboratory.

A practical class, " Robot Creative Practice", in which each student designs and builds an autonomous air hockey robot through repeated trial and error has been proposed. Paper materials were used for structural components which is very easy to redesign in order to be able to instruct approximately 20 students within a limited time and budget. The students conceived and designed the robot by using of the microcomputer and CAD tools that they had learned in the previous year. Their unique robots were reported at a presentation and students evaluated each other. In addition, other related subjects were also provided, resulting in a comprehensive and well-rounded robot engineering education. In this article, the preparation for the opening of the course and the progress of the course after its opening will be described.

The results of questionnaires for participants and internal staff of the IHI Manufacturing Workshop, regional events held at IHI's Yokohama Office and Toyosu Head Office, were analyzed in terms of IHI’s name making better known in the community and IHI’s employee engagement.

We have developed an autonomous mobile robot for learning how to use RTK-GNSS in our educational field. Global navigation satellite system (GNSS) is widely used to track the actual location of vehicles. Recently, real-time kinematic GNSS (RTK-GNSS), which provides up to centimeter-level accuracy, has been spreading in the fields like land surveying, ICT construction, smart agriculture and drone. Considering the technological innovation, we believe that a platform to learn RTK-GNSS is necessary in the educational filed as well. This paper reports the structure of the platform mobile robot. Furthermore, the basic experiment of autonomous driving control is shown.

Papers submitted must be original, and previously unpublished. The responsibility for the contents of published articles rests solely with the authors and not the society. Copyright of the papers published belongs to the JSME (Japan Society of Mechanical Engineers). [Abstract: Times New Roman, 9pt, 100-150words]

In general, underwater robot joints are sealed by waterproof seals such as O-rings and V-rings to prevent exposure of motors, reduction gears, sensors, and other components to water. However, sealing materials have poor sliding properties, which not only reduces the efficiency of motion, but also makes contact work difficult due to the high rigidity of the joint. This paper reports the results of applying a submergible geared electric servomotor without any waterproof seals to an underwater gripper.

Accurate wave measurements are important for atmospheric modeling and climate prediction.Measurement buoys equipped with GPS and IMU sensors are being developed around the world, but there are issues of accuracy and cost. In this research, multiple IMU sensors are mounted on small buoys, and technology is being developed to estimate wave information from the motion of the small buoys.In this paper, we present the relationship between the design parameters of the small buoys and the accuracy of wave estimation, and the interaction when multiple small buoys are coupled.

We have proposed a resolved acceleration control (RAC) method for underwater vehicle-manipulator systems (UVMS) which is a position control method. The model of UVMS used in the control system has modeling errors in the fluid forces acting on the robot. Therefore, we have proposed a RAC with disturbance observer (DOB) in a previous study, and confirmed its effectiveness. Here, a model error compensator (MEC) that is simpler in structure than DOB and specializes in modeling error and disturbance suppression is proposed. In this paper, we propose a control system of RAC with MEC and show the effectiveness of the control system by simulations using a dual arm underwater robot.

To obtain a long-term data of marine resources which are distributed over a wide area of the deep sea, we have been developed a long-term observation system using cable-restricted underwater vehicle. This paper reports about long-term observation system and its wet test performed at 3 m depth. As an evaluation experiment result, the vehicle tethered by a cable was able to cruise with an error of less than 0.34 m against the planned route. The mosaic image generated based on the captured images was able to measure the width, length, and height of the object with an error of less than 4 cm.

For care robots to be used on a daily life, accountability to various stakeholders is important. In many cases, however, interfaces and specifications are only created empirically. Therefore, we have proposed design guidelines for interfaces to achieve accountability to various stakeholders, and user interfaces were developed and validated by experiments. The interfaces are only for the robot users and caregivers, however, there are some other stakeholders for care robots, such as engineers. In this study, a failure investigation interface is developed following the design guidelines, for engineers and legal officials. The effectiveness of the interface is confirmed by validation experiments.

In recent times, there has been extensive research and development of Advanced Driver Assistance Systems (ADAS), enabling drivers to experience an increase in driving safety and convenience. Personalization of such assistance systems is crucial to ensure ease of use for the driver, otherwise, the system may become inconvenient and disabled. Authors constructed a driver model with a dead zone and probabilistic parameters to reproduce the acceptable variation of driving behavior for humans. An evaluation index was proposed to measure the similarity of the distribution of driving behavior, showing the model’s versatility. However, the generality of the obtained model was not verified. This paper tests the robustness of the proposed driver model against the change of the driving condition and the generality of the obtained model.

With the evolution of network infrastructure in the world, the application of robots in daily basis is expected. Even though teleoperation has made many tasks easier to accomplish, teleoperation is affected by network conditions. One example of this is random interruption of camera images. This paper summarizes a proposed delay compensation method using virtual buffer. The effectiveness of proposed method is confirmed on a simple simulation platform. The result shows possibilities in suppressing the fluctuation of the frame-rate of the camera images. But there are still some improvements to be made for future works.

This study examined methods for segmenting phoneme information text for dialect identification with the aim of improving the accuracy of speech recognition of dialectal speech. The results of discriminating dialect data segmented by the three proposed methods were compared in terms of identification accuracy. The result of the comparison showed that the method of splitting the text every 5-gram gave the best accuracy.

In recent years, social robots have been used everywhere, and the office is no exception. The office environment has become more flexible, and there are concerns about the lack of communication and its negative effects on physical and mental health. As a first step to solve this problem, this study develops a system for office robots that can notice changes in health conditions by visualizing emotions. The system displays emotion data acquired by a camera on a graph and a map, and aims to improve the atmosphere of the entire office by linking it with a mobile robot. Demonstration of the proposed system confirmed that office workers can reflect on their emotional state and notice emotional changes.

Semantic SLAM is defined as the extraction and integration of semantic awareness with geometric information to produce dense, multi-layered maps. Semantic maps are crucial for service robots to perform higher-level tasks while fostering human-robot interactions. Searching for objects is one such higher-level task that will become increasingly necessary for service robots. Object search involves the integration of scene-awareness with common-sense knowledge base. In this paper, we propose a method to combine our semantic map with probabilistic priors to find objects in an indoor environment. Our method also incorporates the ability to search for multiple objects during the task. In conjunction with our framework, we present the quantitative results of our experiments in an indoor environment.

The paper describes a prototype mobile manipulator to perform garbage collection tasks. The mobile manipulator consists of a 7 d.o.f. manipulator and an omni-directional mobile mechanism. An RGB-D sensor is mounted on the body for recognition of a target object and measurement of the distance to the object. The mobile manipulator is controlled by a tele-operator with a wireless controller for moving. In manipulation, a target object is recognized and grasped by the manipulator automatically. In the experiment, the mobile manipulator was moved to near a desk by a tele-operator, and the manipulator automatically grasps a target object on the desk.

This paper describes an algorithm for determining the location of dishes in a dishwasher plate rack. We present an mathematical programming problem based on a rectangular packing problem in order to obtain a high packing density in a short period of time. Numerical simulations using a variety of tableware validate the effectiveness of the proposed algorithm.

Sleep quality is closely related to health risks, and its improvement has been particularly sought after in recent years. In addition, the ease of tossing and turning is known to be an important factor in improving sleep quality. However, previous studies have focused on mattresses, which are costly. In this study, we focus on pillows and propose a method to dynamically change their shape during sleep. To verify the effectiveness of this method, we developed a pillow device that implements this method and conducted experiments. Participants in the experiment continuously tossed and turned on the pillow device, which was classified according to their body movements, and their tossing and turning performance was evaluated using electromyography and a questionnaire. According to the results, the participants confirmed an improvement in tossing and turning in each evaluation, with a particularly high improvement in the lower limb-precedence type (p = 0.05). This result suggests that the proposed method can realize continuous improvement of tossability.

This research addresses the growing demand for a modular-based system that enables development in users’ environments, considering security concerns and implementation costs. The study presents a modular approach using the NEP communication library, specifically designed to easily integrate different robotics systems, to enhance gesture-based interaction in remote-controlled humanoid robots in open environments like meeting rooms. The implemented system employs a single camera setup and recognizes upper body gestures, utilizing a Vector robot, a computer with sufficient processing power, and the NEP library for module communication. The modules include environment interaction, person detection, pose recognition, and gesture recognition. Results show a three times increase in frame rate when switching from pipeline to NEP processing and a 20% increase in latency. In conclusion, the modular system using the NEP library successfully improves gesture-based interaction, opening up possibilities for further development and customization in users’ environments.

An escort robot has been proposed to perform with human users cooperatively by observing implicit cues such as eye movement. It is necessary for such robots to plan the path in which the robot does not collide with the escorting person. In this research, we focus on the scenario when the robot moves along a person in challenging environment. This paper proposes a cooperative safe path planning method to establish the robot behavior that avoids colliding with the human by predicting the human’s cooperative behavior. The proposed coordinated behavior is computed based on the iterative route planning taking the human behavior into account. The verification by numerical simulation showed that the proposed method generated safer and more efficient routes than the conventional method.

When a new robot is introduced into a human-activated environment, there is a need for a function that enables users to teach the robot the tasks and motion skills they want it to perform on the spot. Since imitation learning, a typical skill learning method, has the problem of high cost for teaching repetitive actions, we propose a skill learning method that uses data automatically collected by a robot that iteratively executes actions based on a single action instruction from the user. In this study, we have classified the properties of repetitive and automatic data-collection tasks into two: ease of locating and ease of detecting failures. By learning the motion skills of these properties, a series of environmental equipment manipulation tasks can be robustly executed. The effectiveness of the proposed system was confirmed through experiments on actual robot.