We have proposed methods to improve the discrimination accuracy without changing the current anomaly discrimination system. The effect of this method was confirmed for the discrimination system by the MT method used in the actual mass production processing machine. We found that the correct rate is improved by preprocessing the raw acceleration data (time series) that is converted into feature vectors and used for learning and discrimination with an optimized multi-band digital filter. Furthermore, optimizing the cutout range of acceleration data at the same time as the filter parameters greatly contributed to the improvement of the correct rate.

This paper discusses the development of an environment recognition system in an autonomous overhead traveling crane. In the conventional studies, the trajectory planning methods with load sway suppression and obstacles avoidance have been proposed. The planning methods need the environment recognition with the obstacles location and size in the transport space. Therefore, the environment recognition system which enables to construct automatically the transport prohibited areas in crane transport space is proposed in this paper. In the proposed system, the transport space can be detected by a LiDAR, and the detected data are formed in the grid map. The obstacles are recognized by the morphology transform, the labeling process and k-means. The efficacy of the proposed system is verified by the experiments with the laboratory-type overhead traveling crane.

Automation has been adopted and realized in many industrial fields in recent years. However, it has been barely implemented in the field of dealing with living creatures. Therefore in this paper, we propose a robotic system for grasping and transporting jellyfish to automate operations involving living creatures. We created a wire-driven robotic gripper specialized for grasping jellyfish that is very soft with an extremely low friction coefficient. We trained a YOLOv5 model for recognizing jellyfish and obtain the grasping position. In experiments, we used gel jellyfish fabricated using a 3D gel printer instead of living jellyfish. Finally, we conducted an experiment to automatically detect, grasp, and transport jellyfish from one tank to another with the proposed robotic gripper and deep learning based recognition method.

This study proposes the methodology to optimize the motion of a manipulator that equips a pneumatic soft robotic gripper for improving the cycle time. The proposed methodology tunes the motion speed and acceleration of the manipulator, using the Bayesian optimization. The flexibility of the soft gripper could cause in its vibration that prevents the accurate positioning of the manipulator. The slow motion of the manipulator makes the vibration small, however, the cycle times become longer. Hence, this study presents the motion- optimization methodology to reduce the cycle time including the manipulator's operation and vibration-convergence times. Considering that it is difficult to analyze the dynamics of the soft grippe, the motion parameters are optimized through the actual exploration.

In recent years, the demand for cobots that share space with human has been increasing. Since the positions of objects in the shared space are uncertain, it takes time to teach the robot using conventional methods. Therefore, to reduce the teaching time, the external force estimation and admittance control are performed. In this study, as an example of working in a shared space, the robot that opens and closes the door is shown.

To realize the task of stitching the edges of semi-circular and rectangular parts together in three dimensions with a robot arm, we have proposed and studied several methods to feed the edges of the parts accurately including a guide set or a semi-circular robot hand. To investigate the conditions suitable for automatic stitching of the target parts in three dimensions, we conducted some experiments of the combination of different proposed methods. As a result, we were able to realize three-dimensional sewing of two different shaped parts by the robot arm. In this report, we describe the results of verification and evaluation of the proposed methods by experiments.

In the process of producing tulip bulbs, it is very difficult to find and extract diseased strains, so the mechanization of that work has been delayed. Therefore, in order to contribute to the mechanization of this process, we attempted to detect the diseased strain using images of tulip strains being cultivated. Tulip strains cultivated in the open field were photographed at each growth stage for some viral symptoms of some varieties and an image database was created. Using these image data, a discriminator using a convolutional neural network was trained, and the discriminant accuracy was evaluated. As the results, high discrimination accuracy was generally obtained.

This paper presents a method for identifying strawberry plants (fruits, flowers, calyxes, trusses) with the aim of developing a multi-functional agricultural work assist robot for small-scale facility. We focus on semantic segmentation by deep learning. Some pre-trained CNNs (ResNet-18, ResNet-50, Xception and MobileNetV2) are compared as the initial value of weights for feature extraction in DeepLabV3+. In this study, the ResNet-50 is used as the backbone network of DeepLabV3+. In addition, we propose a method of performing post-processing based on the shape characteristics of plants on the results of semantic segmentation. The proposed method is evaluated using the images obtained in the actual strawberry farm, and its accuracy is evaluated by mean IoU. The effectiveness of the proposed method is shown by comparing with and without post-processing. The maximum was 0.731 for fruits and the minimum was 0.294 for trusses (0.643 and 0.199 respectively without post-processing).

Farmer has been aging rapidly in Japan and therefore the workload reduction has been required. One of hard labor is mower. We then have been developing a mower for irregular ground and working by convenient operation for everybody. In this paper, the structure and performance of this device we have been developing are described.

Robots used at vineyards are either autonomous or tele-operated by a remote operator who controls the robot motion based on a live video feed from the robot’s cameras. Unlike structured environments like buildings which have smooth ground surface, the surface of vineyards are uneven. There could be jitters in the video due to a bumpy ground, or loose and vibrating mechanical frame on which the camera has been mounted. Jitters or shakes in these videos are undesired for tele-operation, and to maintain desired quality of service. In this paper, we present an algorithm to stabilize the undesired jitters in a shaky video. The algorithm works by tracking robust feature points in the successive frames of the camera, smoothing the trajectory, and generating desired transformations to output a stabilized video.

As a herbicide-free weeding method, our research group are developing an autonomous running paddy weeding robot that can travel by a simple operation. The paddy weeding robot removes the weeds by churning up the soil and inhibits the growth of the weeds by blocking-off sunlight. Before now, the paddy weeding robot has detected rice seedlings with capacitive touch sensors and turned to next rice seedlings row. However, the detection accuracy to move next rice seedlings row wasn’t sufficient. In this paper, we propose new rice seedlings detection methods. One is method which can detect rice seedlings from images taken by a camera sensor using You Only Lock Once (YOLO). The other is method which can detect the rice plants by a Light Detection and Ranging (LiDAR). The experiment results demonstrate that the proposed rice seedlings detection methods are effective in turning motion for the paddy weeding robot.

In the agricultural field in recent years, unmanned mowing robots that can perform mowing work safely and efficiently are attracting attention as a countermeasure against aging and population decline. Therefore, in this research, we developed a mowing robot that avoids collisions with obstacles by image recognition. Collision avoidance is performed by identifying the type of recognized obstacle, estimating the distance to the target, and combining it with the automatic control of the robot. As a result of experimentally evaluating the collision avoidance performance of the robot according to the type and distance of the obstacle, it was confirmed that the collision avoidance technology with the obstacle by the image recognition developed this time is useful.

The goal of this research is to develop a method to evaluate the quality of chrysanthemum by the degree of stem bending. Criteria for the quality assessment include the state of flowering and the curvature of the stem. This research deals with an image-based quality assessment using the curvature of the stem. In order to measure the curvature of the stem, it is necessary to extract the stem regions from the image of chrysanthemum. Although a previous method used the XYZ color space for stem detection, its accuracy is lower because the colors of the stems and the leaves are similar. In this research, we perform stem detection by an image processing-based and a deep learning-based methods and compare their detection accuracy. We them define a feature that represents the curvature of the stems to judge whether the quality is good. The accuracy of stem detection by deep learning-based method is better than that by the image processing-based one. The accuracy of quality assessment is 0.869.

The aging society has required automatic harvesting equipment in Japan. Spinach is one of the difficult vegetables to harvest by automated equipment. Automated harvesters using the passive handling method can manage a wide range of soil and species of spinach. The soil reaction force acting on its cutting blade is crucial for designing its cutting path and equipment. However, theoretical consideration for the relationship between the blade and reaction force from the soil has remained insufficient. Therefore, this paper constructs the theoretical model using the resistive force theory including the shear force. Then, this paper estimates the reaction force of the cutting blade. The results showed that the constructed model captures the experimental trend of the reaction force according to its arm angle. Furthermore, the error rate between the experiment and the estimated result was less than approximately 20%

Robotic weeding manipulation must remove weeds without damaging crops. To achieve this, the robot must be able to detect individual weeds in the image of fields where crops and weeds are mixed and operate the weeding device on those coordinates. We develop a system to detect weeds using Mask-RCNN algorithm from the image of ridges captured by Intel Realsense, and evaluate the effect of the annotation shape of the training data used in this algorithm. We compare the performances of square and polygonal segmentations. Results of comparing based on a single image, accuracy of polygonal segmentation is better than that of square segmentation. In addition, polygonal segmentation has a segmentation mask which each detected weeds along the contours.

The number of agricultural workers is declining. To solve this problem, smart agriculture that utilizes advanced technologies such as robots, AI, and IoT is expected. If the harvesting work can be automated, it can be expected to reduce the burden on agricultural workers. In this study, tomatoes were set as the crops to be harvested. Red area recognition and circle-shaped recognition are important factors in tomato recognition. The most severe condition in this light environment is a situation called backlight, in which the sun is in front of the camera. When a lot of sunlight enters the camera, it causes overexposure, and it is not possible to find not only the red color of the tomato but also the shape of a circle. We developed a camera stand that reduces the influence of backlight so that the target tomato can be recognized even in such a light environment.

Wire-driven continuum robots exhibit great capabilities for manipulation and locomotion in complex and unstructured environments due to their inherent compliance, flexibility, and dexterity. This paper proposes a wire-driven exoskeleton continuum robot which has high rigidity and can exhibit hyper multi-DOF. First, the structure and actuation design of the proposed exoskeleton continuum robot is discussed. Then the control system is presented along with the application of deep learning in creating a model to find the relationship between the wire pull-lengths and robot arm posture coordinates. Finally, an experiment was carried out using that deep learning model and the results were discussed. The accuracy from the experiment was found to be 10.22 mm in average.

It is necessary to satisfy the knowledge that contributes to their well-being based on the background of the increase in construction workers' labor burden. Therefore, the purpose of this study was to clarify the characteristics of self-control and the expression of ego depletion based on the experiences of construction engineers by qualitative research methods. As a result, working behavior (e.g., consideration of safety, adjustment of construction work's schedule) and control of negative emotion (e.g., impatience and interpersonal irritation) were extracted as their self-control. It was also clarified that cognitive and behavioral self-control necessary for the work became impossible (e.g., failure to decide work processes, lack of attention), and negative emotion appeared as ego depletion. Future research should aim to reveal the relationships among the specific variables extracted in this study and to examine practical techniques for intervening in appropriate self-control and ego depletion to contribute to improving well-being.

The article reports evaluation results of human dynamics as a controller in order to improve operability of positioning control systems of a crane load, including anti-sway control. Specifically, we evaluated the human dynamics as a controller using a task in which a human follows a suspended load to a randomly changing target position as an example. The results showed that when there is no anti-sway control of the load, it has a characteristic including a notch filter so that human can damp the swaying of the load, but when the anti-sway control is implemented, it has a proportional characteristic. In other words, it is thought that a crane suspension control system that is easy for humans to operate can be constructed by introducing appropriate anti-sway control.

We have developed three types of tunnel inspection system with combination of robotics, an acoustic analysis, and an image procession. These systems have robotic hammers and light-section cameras on a work platform to inspect tunnel lining concrete including zenith parts which are difficult to be inspected. Although these systems require a crane and an assembly location nearby the tunnel. So we developed a new system with swing and telescopic arm mounted on an aerial work vehicle, which can be used to inspect wall efficiently while the vehicle is running. The system does not require on-site assembly work and can equip the same hammer and camera of the ordinary type. In this report, we introduce the outline of the new system and operations tests in an actual tunnel inspection.

In order to patrol inspection points, a patrol path planning, which plans both a route though the inspection points and paths between the inspection points, is required. The planning has a problem that the more inspection points, the more planning time. We focused on path planning occupies most of the time for patrol route planning, we apply lower bound costs for solving traveling salesman problem. Approximation with the lower bound reduces the number of path planning, which reduces route planning time. The simulation with 25 points inspection showed that the planning time decreased 94% in comparison with the conventional method, whereas route length increased only 0.6%.

In unmanned construction, accurate spatial cognition by the operator is difficult. In order to solve this problem, we focused on the operator's spatial recognition and moving a viewpoint of drone camera. Since the motion parallax, which is one of depth cues obtained from images, is reproduced by active viewpoint movement, we used head movement to move the viewpoint. We have set Manual Method for operating the viewpoint position and Semi-Automatic Method for operating only the direction and timing of the viewpoint movement as the operation method for moving the viewpoint. Therefore, we analyzed the effects of difference in methods of moving a viewpoint on the operator's spatial cognition, using tasks that do not depend on the operation skills of construction machine.

We developed a rope-type climbing robot that enables spiral climbing for the purpose of infrastructure inspection. This robot has a spiral propulsion mechanism that controls the attitude by tilting the roller to the tether. We developed a two-wheel spiral climbing robot equipped with a wheel drop prevention plate. The wheel drop prevention plate has the function of preventing the wheels from falling off. Then, we configured optimal synchronous control system of the two rollers to reduce the load on the motor. The attitude of the spiral climbing robot was controlled by PID control, and it was confirmed that the robot followed the target attitude.

For infrastructure equipment such as power plants and tunnels, automation and labor saving of patrol inspections are important issues. So far, humans perform inspection even in dangerous sites or at high places. Recently, inspections using drone have been carried out, but it is necessary for a skilled pilot to manually operate it remotely. We have developed a system that focused on patrols of inspection points in a tunnel. It specifies inspection points, creates flight routes from 3D models, and send flight commands from robot management systems to drone. This system eliminates a skilled pilot, allowing the drone to move on a safe flight route and approaching the inspection site to take high-quality inspection images.

We have examined a pipe traveling robot using microwave guided-modes propagating along PVC pipe-wall. It was confirmed that the effect of the guide that restricted the flexibility with the spring which produced a pipe traveling robot.

In the Japanese construction industry, there is a need to improve productivity per capita due to problems such as a declining working population and an aging population. Therefore, the development of teleoperated hydraulic excavators is being promoted. However, the work efficiency of teleoperated hydraulic excavators may decrease. One of the reasons is that it is difficult to see the surrounding environment, and the introduction of a bird’s-eye view is being considered to solve this problem. However, work efficiency may decrease due to the influence of mental rotation. The purpose of this study is to develop an interface for a teleoperated hydraulic excavator that can reduce the cognitive load caused by mental rotation. As a result of this verification, it was confirmed that the operation time was significantly reduced under specific conditions. In addition, there was no significant difference in the Sense of Agency.

In this research, we develop a two-wheeled wire traction vehicle to support the automation of slope inspection. By using the information from the tension sensor and the acceleration sensor, we have shown that the vehicle can move on the slope. Then we have shown methods that the vehicle overcome the slope with steps, and step over the steps on slope with a height of 200mm. In addition, we propose what type of input needs to further automate the system and how to get input. Then we also propose how to enable it to move in a two-dimensional plane field.

In recent years, the shortage of successors at construction sites has become more serious. At the same time, autonomous driving technology for automobiles at lower prices for LiDAR is progressing. At this background, this research aims to realizing an automating crawler crane. However, its verification test is not easily carried out with an actual machine due to cost, difficulty in ensuring safety, and human and mechanical loads. Instead, we build a crawler crane simulator model on Gazebo. This paper introduces our crane model on the Gazebo simulator, hoping to perform SLAM with the crane model.

At construction sites, there is a need to develop systems that automate simple tasks. Therefore, this study focused on the crushing of rubble generated during building demolition. A rubble recognition system was developed to recognize and identify target rubble for automatic crushing work by construction robots.

This paper presents a video-based active sensing method that can monitor structural anomalies such as cracks by inspecting structural parameter changes with digital image correlation analysis when a load is operationally shifting on a structure to be analyzed. Several experiments were conducted for structural anomalies visualization on an aluminum cantilever beam of 2-meter length by analyzing sub-pixel displacement distributions on many points for high-resolution video sequences at hundreds of frames per second. The effectiveness of our concept was verified by showing the temporal changes of vibration spectra with specific peak frequencies when a load of 100-gram weight was passing around the artificially-made crack of the cantilever beam.

In this study, we develop a sensor terminal with multiple and various sensors named sensor pod, which collects various environmental information at a construction site. The sensor pod is equipped with a 3D-LiDAR and a vibration sensor, which can be used to predict the surrounding hazards and evaluate the ground stiffness. In this paper, we introduce a method of evaluating ground stiffness using the waveform distortion of multi-point synchronized vibration data obtained by the distributed sensor pods.

In this study, we propose an algorithm for planning the bucket edge trajectory to automate the operation of the backhoe to excavate the ground. In proposed algorithm, the shape of the ground before start excavating operation is obtained based on point cloud data using 3D LiDAR and used for trajectory planning. This trajectory planning takes into account that the bucket is easy to penetrate at the start of excavation, the soil has the target volume at the end of excavation, and the soil does not easily spill from the bucket after excavation. The proposed method was implemented on an actual backhoe and experimented in the test field of Public Works Research Institute. These are described in this paper.

We proposed a model in which three small cranes are used to form a three-dimensional truss structure to perform construction work in a coordinated manner. In this model, the workspace and payload depend only on the strength of the structure because the structure does not tip over as in conventional cranes and excavators. This makes it possible to lift loads while maintaining large workspace, despite the overall light weight. We have also proposed an assembly method to construct the 3D truss structure. In order to verify the feasibility of the proposed model, we conducted small-scale operation verification using radio control models. The operation of the prototype model was confirmed to be highly feasible in all processes of assembly, transportation the bucket, and excavation.

The paper reports a method to improve the accuracy of radiation thermometry by infrared thermo-camera equipped on a small UAV (Unmanned Aerial Vehicle), eliminating the disturbances in the outdoor environment. In this study, the authors are developing a system for detecting anomalies associated with the deterioration of piping equipment in a geothermal power plant. However, it was challenging to obtain accuracy with an infrared thermo-camera due to the significant influence of external radiation disturbance such as solar radiation and sky reflection in an outdoor environment. We proposed models of compensating for those disturbances. Those improved the accuracy to detect anomalies of the facility by considering the characteristics of emissivity and reflected radiation in the measurement. Consequently, by thermal model corresponding to the effect of solar radiation and sky reflection, we confirmed that the infrared thermo-camera on UAV could accurately detect temperature anomalies in plant facilities, even in an outdoor environment.

The initiative of Digital Transformation (DX) in the field of infrastructure has been more and more active. In the DX, research and development of autonomous construction machinery, especially autonomous excavators, has been carried out with the aim of improving productivity. In excavation of soil, the motion of the hydraulic excavator is affected by the characteristics of the soil. Therefore, it is necessary to apply control system in consideration of the characteristics of the work object to the autonomous excavator that can realize desired operation at various sites. On the other hand, Database-Driven Model Predictive Control (DD-MPC) considering the change of the controlled object has been developed by authors. This paper presents the method to apply DD-MPC system to the autonomous excavator. The effectiveness of the proposed method is verified by a simulation.

We examined the possibility of managing construction by a tele-existence robot. We prototyped a robot with basic functions of moving around, taking pictures, and implementing functional safety measures for construction sites. We confirmed the needs from the results of application experiments at multiple construction sites. In addition, through co-creation with the construction site manager, we materialized the functions and how to use them.

In a geothermal power plant, the soundness of the pipeline is an essential factor that could influence power generation efficiency. However, it is laborious and dangerous for staff to inspect the condition of the pipeline. Therefore, we have developed a system that can conduct inspections automatically by using a small unmanned aerial vehicle (UAV) equipped with Laser Imaging Detection and Ranging (LiDAR). In this research, we focus on increasing point clouds' accuracy and improving the efficiency of inspection route generation. The experimental results show that our system could detect less than 25 mm misalignment and estimate UAV attitude less than 0.05 deg. Furthermore, the efficient inspection route could improve abnormalities detection accuracy from 66.8% to 75.8%.

The robot that climbs a belt with a pressing force adjustment system have been developed and studied to examine the load on the motor and the slippage of the rollers in terms of electricity consumption. From previous studies, we knew that climbers started without slipping would need to push the rollers excessively, but on the other hand, pushing the rollers too hard would overload the motor during ascent and slow down the robot. We found that controlling the pressing force improved the electricity consumption more than the constant pressing force. Additionally, by adopting that mechanism, we have obtained data on climbing with various constant pressing forces and analyze the behavior and these properties.

Over the past few years, the concern over manufacturing of building materials using a WAAM-based metal 3D printer has risen. Although studies have been conducted mainly with steel-based materials, there are relatively few examples of using aluminum, which is frequently used as exterior materials.

In order to investigate the feasibility of using the metal 3D printer for large aluminum components, a chair was designed and manufactured. In the design, the chair with a hollow cross-section of 420 mm (width) x 400 mm (depth) x 600 mm (height) was designed. In manufacturing, it was confirmed that the designed chair could be fabricated, indicating that a relatively large hollow structure made of aluminum can be fabricated with the WAAM-based metal 3D printer.

Currently, the number of farmers in Japan is aging. In addition, the epidemic of the new coronavirus has raised the demand for infection control measures. Therefore, there is a need for efficient production with fewer workers. In the case of plastic greenhouses, the physical burden on employees is high because they must work in a hot and humid environment while wearing a mask to prevent infection. To reduce the workload of the employees, we propose to use robots to transport the harvested crops. This system can reduce the workload of transporting heavy materials and reduce the risk of infection among employees. In this paper, we propose a robot to transport harvested crops in a plastic greenhouse.

In response to the growing needs for labor saving in distribution centers, the demand for automation of work is increasing. However, due to the variety of items, it is difficult to automate with conventional technology. In order to automate the picking work of various items, we have developed a picking robot system equipped with item recognition technology, motion planning technology, and hybrid hand. In this paper, we show results of evaluation experiments on various items and how they can be handled.

The spread of the novel coronavirus (COVID-19) has revealed issues regarding the number of polymerase chain reaction (PCR) tests that can be conducted daily, testing system used, and organizations that conduct these tests in Japan. This research aims to realize an automated inspection system by utilizing robotics technology and information management technology to prepare for possible future virus mutations and pandemics. In this paper, the requirements and technology trend surveys and issues about PCR inspections are extracted and a new concept is formed from the results. The functionality of the system consists of the following processes: pre-stage process, RNA extraction, real-time PCR, and information aggregation. Furthermore, the system is flexible and continuous. This system is particularly effective in prefectural public health institutes, which play an important role in the early stages of a pandemic. Future works of this study may consider the enhancement of the pre-stage and information aggregation processes of the system.

The spread of the novel coronavirus (COVID-19) has revealed issues regarding the number of polymerase chain reaction (PCR) tests that can be conducted per day and the testing system and organization in Japan. This research aims to realize a effective testing system to prepare for future virus mutations. In this study, we propose a microtube capper as an element of the technology development of an effective pre-stage processing system. The proposed microtube capper consists of two arms that can operate independently to open and close caps of microtubes. The experimental results showed that the proposed microtube capper could smoothly open and close caps of different types of 1.5 mL microtubes. Future studies may consider integrating the proposed microtube capper into the pre-stage processing robotic system for total evaluation.

The spread of the new coronavirus (COVID-19) has revealed that only a small number of polymerase chain reaction (PCR) tests can be performed per day in Japan. This study aims to develop a novel, flexible, and continuous testing system for new mutant strains and unknown viruses that may occur in the future. In this paper, we have analyzed the current problems related to the information system overseeing PCR testing, described the ideal system configuration, and designed and prototyped an inspection information system. The experimental results showed that the proposed inspection information system could import/export inspection information and manage the inspection process. Future studies may be performed to integrate the proposed inspection information system into the inspection process for a more comprehensive evaluation.

This paper describes the fundamental experiment for inspecting the paint coating thickness of spline plates of bridges with a teleoperated manipulator. In general, unstable suspended scaffolds are used at the bridge construction sites. In this study, the unstable scaffolds and the inspection area of a bridge are imitated, and the feasibility of the coating thickness inspection with a teleoperated manipulator is evaluated. The experimental results showed the possibility of the inspection remotely. Besides, the results clarified the issues that the robot swung back and forth during the operation due to the unstable scaffolds and the difficulty of pointing to the same inspection point due to its swing.

Since 1988, the Department of Electronic Control System Engineering at National Institute of Technology, Numazu College has conducted development of Micro Intelligent Robot System“ MIRS ” as a Project-Based Learning (PBL) into the curriculum for the education of system development. The purpose of this project is to provide a place to learn ”the process of system development,” ”education as creative education,” and ”social implementation experiments” through robot development. In this paper, we report on the results of the socially-implemented robot education conducted under the theme of ”Life with Robots. The analysis of the past five years of practice results in this paper shows that the teams that conducted social implementation experiments had higher technology evaluation scores than those that did not. However, even among the groups that did not set a theme for possible social implementation, there were teams with high technology evaluation scores. Furthermore, because of the emphasis on social performance, groups with low technology evaluation scores appeared, indicating that it is necessary to set themes that allow social implementation and technological achievement.

While the automation of overhead travelling cranes is progressing, the ‘dynamic lift off’ process is still manually operated. In this paper, we thus developed an automatic ‘dynamic lift off ’system based on a trolly-position control system using wire tension. The system suppresses swaying of the suspended load after the dynamic lift off by correcting the suspension position so that the combined direction of the tension of each suspension wire matches the gravity direction of the suspended load. As a result of the experiment using the physical simulation using Unity and the developed scale model, we confirmed that the sway of the suspended load can be suppressed by using the automatic trolly-position control system than the manual operation.

The task-board task is an assembly task for evaluating the ability of a robotic assembly system. In the World Robot Summit (WRS) 2018, the task-board task was performed with robots. Acknowledging that the competition aims at jig-less assembly of multiple parts without setup changes, in this paper, we analyze the performance of robots in the top four participating teams in the competition with respect to usage of tools (i.e., time of using or exchanging tools and the number of tools) and investigate the relation with the whole work time.

In recent years, many researchers have been working on a connector insertion task using a real robot. There is a wide variety of connectors used at the manufacturing site, and it is necessary to insert them while changing gripping them. While there are many research results that are robust against socket misalignment, there are few research results that are robust against connector misalignment. In this research, we propose a system that estimates the grip position deviation from the image taken by the optical tactile sensor and corrects the demo trajectory. We also applied this system to connector insertion in a real robot environment and evaluated its performance.