We have built an intelligent robot reference architecture with ROS. Using this architecture as a platform, the development efficiency of picking robots and unloading robots has improved. The modularization and core assetization of robot functions have made it easy to add or change functions and reuse software in the future. In addition, the platform based on ROS has made it possible to easily incorporate open technologies through open innovation.

The evolution of robotics technologies has produced many kinds of robots, for example, cleaning robots, guidance robots, and transfer robots. In the future, these robots will coexist in the same public facilities. However, we will not be able to avoid developing these robot systems inefficiently, because most robot products have different interface specifications from each other. To efficiently develop robot systems, RRI (Robot Revolution & Industrial IoT Initiative)’s subcommittee has defined interoperable mobility function interface specification and published it in 2020. In this paper, to extend the specification document, practical systems operating conditions are discussed, and more detailed standard behavior and life cycle of mobile robots system included interface definitions are proposed.

Autonomous mobile robots are increasingly deployed in manufacturing and logistics, as well as in public facilities, where the services they perform include delivery, security, and cleaning. However, there are still obstacles holding back their use. One of the biggest is getting different kinds of robots from different manufacturers to work together. Robots from different manufacturers cannot be connected with the same operations control system, and introducing a new mobile robot to work alongside robots from another manufacturer requires redesign of the system. Similarly, companies use their own maps to move their robots.

Robot Revolution and Industrial IoT Initiative (RRI), Robot Innovation WG (WG3), Software Architecture Examination Committee has been studying for the standardization of mobile robot interfaces since 2018. In this presentation, we will introduce the map specifications (Localization Map, Environment Map, and Operation Management Map) and their surrounding architecture.

Papers submitted must be original, and previously unpublished.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).

With the increasing focus on service robots due to the labor shortage, we propose a flexible mobility control system by combining Node-RED as the upper system and ROS and RTM as the robot mobility control. In addition, we use odometry as a countermeasure against wandering in SLAM and AR marker tracking by RTM to correct the path. In the experiments, the validity of the proposed method was confirmed.

This paper introduces the human resource development course and its activities that are now open to the NEDO special course project that started in June 2020. The purpose of this project is to promote the dissemination of the results of the NEDO "Technology Development Project for Robot Commercialization Applications" project, which ended in March 2020, and to develop human resources who can utilize the results. Currently, we are offering online lectures on ROS related manipulation using 3D sensors, AI application seminars, and symposiums using the results of the project. In this report, we present the status of these activities and our plans for FY2021.

We are currently developing an Operation-based middleware (OBM) model. This paper describes a sample design model of the mobile robot navigation system using the OBM. In OBM, input-output functions without any states are focused as the unit modules of the distributed software system development for robots. In order to confirm the sufficiency of the OBM model, we designed a mobile robot navigation system. As the result, the needs for the generic functionalities of switching the execution of operations are found out.

To improve the procedure of robot system integration, robot middleware is one of the important solution. However, there are few reports and theories how to develop the software modules in robot middleware. To be advance the robot software development, standardized robot software modules are required. Based on this motivation, this paper introduces the activity of the“ Robot Software Architecture Working Group ”established in RRI, which includes the specification of mobile robot interface and manipulation interface.

In recent years, not only industrial robots that work in factories but also service robots that substitute work in the human living environment are becoming widespread. However, the development of service robots has to deal with diverse and complicated living environments of people, and there is a problem that the development becomes large-scale. Therefore, instead of developing a robot from scratch, a mechanism called middleware for robots was developed that can reuse the elemental technologies of the robot. However, there is no design guideline, and the robot software module is made by the developer's sense, and the evaluation by word of mouth is the center of reusability. In other words, the problem is that there is no quantitative evaluation guideline for evaluating robot software modules. Therefore, in this research, we proposed an evaluation method for robot software, and applied and considered the evaluation method.

Open-source software (OSS) is increasingly being used in robot development. On the other hand, various issues peculiar to robot development using OSS have also been pointed out. As an initiative to resolve these issues, the RRI (Robot Revolution & Industrial IoT Initiative Council) has set up multiple committees called the Survey and Review Committee as a forum for gathering robot users, SIers, and manufacturers and discussing since FY2018. In addition to these committees, a new committee will be established for the purpose of maintaining and managing OSS for robots and platform software, and the background and purpose of the establishment will be described.

In this paper, we conducted a basic research to develop a system that can improve concentration in simple repetitive practice by visualizing passing techniques in soccer. In the proposed method, we developed a device that can measure the impact position of the foot, and verified the effectiveness of the device because it can classify basic movements such as walking and running and kicking. In addition, by measuring the impact position of the foot using the device, a correlation between the impact position and the accuracy of the pass was observed, but it was found that the information of the impact position of the foot alone was not sufficient.

Understanding the pedaling motion of a bicycle is important for improving pedaling efficiency, preventing injuries, and developing bicycle equipment. Measurement of muscle activity using surface EMG is useful for understanding pedaling behavior. However, when measuring surface EMG in the field, it is necessary to develop smaller sensors and clothing that does not interfere with the rider's movement. In this study, we developed a surface EMG measurement system that can be used in the field. We also conducted bicycle measurements on an actual road field and proposed a method to quantitatively evaluate pedaling motion from power data and surface EMG.

In this paper, we present a method to extract the characteristic of the upper body dynamics in a running motion. Recent progress in the motion measurement technologies including the IMU-based motion capture suit allows us to visualize and evaluate detailed information of biomechanics analysis. It is getting easier for general people to use such applications. In the process to commercialize an IMU-based motion capture suit, it is important to reduce a burden when a user wears a whole-body suit. If we can estimate the whole-body dynamics only from the lower-body measurement, a lower-body suit is enough for the measurement, which can reduce the user’s burden. The purpose of this study is to estimate the whole-body dynamics estimating the upper-body dynamics. We analyze the centroidal dynamics of the upper-body in a running motion, and discuss the extraction of its characteristic.

The aim of this study is to develop an inexpensive and easy-to-use motion measurement system which can assess complex motions, e.g. squat exercise, in non-contact manner. To measure 3D posture of a human being, a compact RGB-D camera is used. Here, 2D human-pose is estimated by a skeleton tracking algorithm using RGB image, and 3D human-pose is reconstructed by mapping the 2D pose and the depth image. In the experiment, we measured center of pressure (CoP) and force output in squat motion using the proposed system. As the result, the proposed system has a potential to assess squat exercise, but further improvements are required for practical use.

We approach from the two viewpoints of spatial accuracy and temporal accuracy, and our main goal is to improve the batting ability of beginners and to enable batting in actual battles. It is considered difficult for beginners to improve their abilities for environmental reasons. If a training method using a VR system is established, this can be resolved and sports can be enjoyed more easily. In addition, since visual sports training using a VR system has not been sufficiently researched, we believe that it can be used as a new sports vision training by verifying its effects and finding its usefulness. Based on these facts, this research focuses on spatial accuracy from the two perspectives of temporal accuracy and spatial accuracy, and aims to develop a system that improves the ability of the batting eye using VR technology. To do.

In recent years, ground golf has become widespread as a place where elderly people can interact with friends while moving their bodies. In this paper we develop and evaluate ground golf system which enable the remote participation in the real ground golf community. This system presents each other ’s live image and voices to remote participants and field participants. In addition, the ball is launched at the same time as the swing of the remote participant. This system enables a new remote communication method. We actually used this system and evaluated it by various methods to clarify the points to be improved.

This study presents a wearable device to support a trainee to learn throwing motion by visualizing fast and subtle movements of finger and wrist. To learn a new complex motion, a trainer tells a trainee how to move the body to reduce the error between ideal and actual motion. The device measures the changes in the internal structure and the external shape of the wrist caused by hand motions by means of the electrical impedance measurement. The measured impedance values are sent to the external terminal that shows its temporal change graphically in real time. As the experiment, the electrical impedance was measured while the basketball jump shots, and the results show the potential of the proposed device as a motion learning support system.

Pedaling skill has been considered as one of the important factors to maximize riding performance of a cyclist. In order to understand the pedaling skills, many studies have used surface EMG to estimate muscular activities of a test subject. However, the measurement of surface EMG requires skin pretreatment and complex attachment works, that becomes burdens of both measurers and test subjects in an experiment. Therefore, the reproducibility of the surface EMG measured on different days is questionable. Then, we look for an alternative to EMG to analyze the features of pedaling exercise of a cyclist.

In this paper, we focused on the sound pressure around a clank during pedaling exercise, and confirmed Granger causality between the measured sound pressure data and the surface EMG.

Stooped posture of the trunk during baseball pitching is able to cause shoulder injury. Although the rectus abdominis and erector spinae muscles significantly contribute to maintain the trunk posture, how activities of those muscles are related to the stooped posture is unknown. Therefore, the purpose of the present study was to determine whether the stooped posture was caused by the trunk muscles during the pitching motion. Three participates performed pitching in the stooped and upright postures. The electromyography (EMG) activity of the rectus abdominis and erector spinae muscles were measured. As a result, there was no difference in the rectus abdominis EMG activity between the stooped and upright conditions, and the erector spinae EMG activity was increased. These results suggest that the stooped posture during pitching was caused by other factors, such as abdominal pressure and the activity of the related muscles.

Recently, the flexible transport system which uses coordinated transport robots will be focused in many applications. In a previous research, the concept and the theory of coordinated transportation had been demonstrated in many studies, but the application to complex path had not been evaluated yet. For the coordinated transportation, robots switch grasping and releasing of the object are required when the formation is changed. In this paper, we propose a cooperative transport system using multiple robots with suction cups to make soft and unloosen grasping. The effectiveness of the transport system is evaluated by fundamental transport experiments.

In recent years, logistics robots that work at logistics warehouse is in the spotlight. In order to improve the work efficiency of the logistics robots, it is important that the logistics robot does not get crowded between the shelves. In this paper, we proposed avoidance control between shelves of the logistics robots and verified it with a simulator.

In recent years, Japan's declining birthrate and aging population have become serious. These problems have a great influence on agriculture, and the labor shortage of agriculture is becoming serious. Therefore, we are developing an agricultural task support robot system as one of the solutions to these problems. This paper focuses on task allocation method for transportation tasks of harvest crops by swarm robots. Based on the task allocation model of red harvester ant (Pogonomyrmex barbatus) and optimal foraging theory of chick, we propose a cooperative conveyance method in which robots autonomously acquisition and perform conveyance tasks that occur scattered and sporadically. Optimal foraging theory is used to calculate the activity required for the division of labor behavior model of ants. The usefulness of the proposed method is verified by simple transportation task simulation.

Most of the welding robots are either large and non-mobile manipulators or small vehicles that do not have an autonomous navigation function. Therefore, they cannot be used for long distance welding in narrow or high places. For this problem, this paper proposes a swarm robot system for automating long distance welding in narrow or high places. The authors have also developed a simulator that consists of a multiple-links cable model that reproduces the friction of surface, elasticity and viscosity in the bending direction of a real cable, and an omni-directional vehicle model that simulates an omni-wheel with zero-friction in the axial direction.

The authors have been developing a swarm robot system to automate long distance welding in narrow and high places, which cannot be handled by existing welding robots. In welding operations, it is necessary to tow not only the welding torch but also the cables that supply welding wire and gas to the torch. In addition, when welding in a narrow space, it is necessary to avoid excessive stress on the cable and to avoid obstacles such as walls. In this paper, the authors propose a method for path planning and autonomous navigation by solving a multi-objective optimization problem that satisfies these conditions. In this paper, the authors consider the case of towing cable on a two-dimensional plane and evaluate the proposed method by simulation.

The mechanical behavior of a rod suspended by the wires attached to multiple manipulators are analyzed when it is pulled out from a hole in a thin plate. The conditions for wedging phenomenon are clarified by the statics calculations with various hole diameter D, plate thickness t, and rod diameter d.

We propose the development of a coverage flight planner that solves the domain coverage problem for the use of multicopters in wide-area surveys and other applications. In such multicopter application cases, collision-free and efficient coverage flight path planning is necessary. In this paper, two approaches to domain decomposition are separated and their optimality is discussed. In addition, a system is developed and demonstrated to set a flight path that is sub-optimal for each region. As a result, each of the two region partitions showed a good evenly divided region, and the flight paths designed on the two region partitions were also efficiently generated. In addition, flight tests of two multicopters were conducted using the generated flight paths.

In this study, we propose a wide area exploration system that combines wire towing system and passive mobile robots. Passive mobile robots loaded with exploration sensors are driven by a pulling force from wire towing system and can move to desired location by steering control. In this paper, we expand the search range by increasing the number the towing units, derived the explorable range that the passive mobile robot can be explored, and verify the explorable range by experiments. In addition, we propose a method of round-trip trajectory and verify its effectiveness by experiments.

In recent years, the importance of autonomous mobile robots at the time of disaster has increased. It is necessary to improve disaster robot technology to carry out quick and safe rescue activities. The mainstream of autonomous mobile robots is multiple cameras, but if the technology for collecting information with a single camera can be improved, the cost can be reduced by reducing the number of cameras installed, and even if the same amount of cameras are installed, it will be profitable, and it is expected that the amount and accuracy of the information received will be improved. Therefore, In this paper, an algorithm that reads camera information from a LEGO Mindstorm equipped with a line laser and a single camera and can measure the straight-line distance to an obstacle in real time even in a dynamic environment by using OpenCV and python together. By using Python and OpenCV, it will be possible to measure the distance to obstacles in real time, and it is thought that dynamic detection will be possible in the future.

In order for robots to operate safely, it is important to sense the environment. However, creating a sensing system from scratch is costly. To improve this situation, we use ROS2. In addition, we thought that a small mobile robot can move safely if it can grasp the space when sensing. Therefore, we use the Intel RealSense Depth Camera D435, which is a depth camera, and the Raspberry Pi 4, which is small and easy to integrate with a depth camera. Therefore, in this paper, we use the latest version of ROS2 Foxy Fitzroy, and Intel RealSense Depth Camera D435, to a Raspberry Pi 4 with 64bit version Ubuntu 20.04LTS installed, and measure the environment with the connected depth camera. We will measure the environment with the connected depth camera.

Recently, many public facilities have been promoted to be barrier-free. In such a situation, an accessibility map that indicates areas that are drivable and areas that are not drivable, such as steps and slopes is very effective for wheelchairs and strollers. This research focuses on viewpoint planning for autonomous 3D measurement by a mobile robot in order to create an accessibility map. For efficient 3D measurements, the viewpoint planning should be done to reduce the size of unknown regions. We plan viewpoints based on the frontier, which is the boundary between the known and the unknown regions. We verify the effectiveness of the proposed method in simulation and by a real robot environment.

The goal of this research is to develop a system that recognizes the surrounding environment by the ultrasonic sensor in high resolution. In previous research, we proposed the ultrasonic sensor system consists of an ultrasonic phased array transmitter and a microphone array with MUSIC method. In this paper, the experiments are conducted under the conditions that the five objects are at the same distance and at different distances. Experimental results show that it is possible to estimate the object positions with an angular resolution of at least 20 deg. Furthermore, we confirmed that it is effective to vary the threshold exponentially for determining the number of estimated objects according to the distance of each object.

The purpose of this research is to evaluate the difficulty of an environment that we want to autonomous mobilities to drive only from prior information. We think that to be enable to evaluate the difficulty would contribute to reducing the risk and the number of times autonomous mobilities are required to perform actual driving tests. We propose a method to evaluate the difficulty by using the achievement rate of the mobile task. This rate is calculated by a function with environmental factors as variables that affect autonomous driving. Therefore, we enable to evaluate the difficulty only from prior information. We also analyzed which factors influenced this difficulty by examining the magnitude of each factor’s effect on the estimation of the mobile task achievement rate.

Our research group aims to develop autonomous mobile robots for urban areas. In particular, we propose a method for autonomous mobile robots using braille blocks installed in urban areas. In addition, we use deep learning method YOLO in braille blocks detection. In this paper, we state the method about grouping braille blocks by the braille block detector output. We have developed a method to discriminate adjacent braille blocks from the overlapping rectangles of the detection results in order to grouping braille blocks. As a result, we were able to correctly group the installed braille blocks for each edges and nodes.

Our purpose of this study is to develop a person following robot using a LRF and a RGB-D sensor on a pan/tilt-rotated mechanism. To acquire the position of the target to be followed, the target must be kept within the sensor’s measurement range. Therefore, in this study, we combine both the methods of recognizing a person in the image by letting it orient the RGB-D sensor on a pan/tilt-rotated mechanism to the target direction and tracking the target of a person with an LRF-based particle filter. In controlling the robot, we have used the DWA method to avoid obstacles and the PI control for its direction change. The experimental results conducted with the combined method will be demonstrated for the effectiveness to the following robot system.

In this paper, we propose a moving object point cloud detection method using point cloud clustering. Conventionally, point clouds obtained with 3D-LiDAR are placed in discretized spaces called occupancy grids, and the static and dynamic environments are estimated by determining whether points exist in each space over time. However, this method misrecognizes parts of the object as partially static, because parts such as the back legs of a pedestrian tend to remain and continue in the occupied grid. Therefore, we overcome the conventional problem by clustering the point clouds obtained with 3D-LiDAR using Depth Map and making static and dynamic judgments for each cluster based on the estimation results of the occupancy grid.

In this research, we develop a garbage collecting robot for marine microplastics. This system uses 3D LiDAR to detect microplastics on a beach and collects them by avoiding obstacles and sea automatically. This paper describes the results of detection experiments of microplastics using reflectance information taken by 3D LiDAR.

The Nakanoshima Challenge is a contest for developing sophisticated navigation systems of robots for collecting garbage in outdoor public spaces. In this study, a robot named Navit(oo)n is designed, and its performance is in public spaces such as city parks evaluated. Navit(oo)n contains two 2D LiDAR scanners with uniaxial gimbal mechanism, improving self-localization robustness on a slope. The gimbal mechanism adjusts the angle of the LiDAR scanner, preventing erroneous ground detection. We evaluate the navigation performance of Navit(oo)n in the Nakanoshima and its Extra Challenges. We evaluate the navigation performance of Navit(oo)n in the Nakanoshima and its Extra Challenges. We also impremented the loop-closure method based on the map switching.

The purpose of this research is to estimate the direction of travel in outdoor passages directly from an image for navigating develop a method to a robot. While humans can define the pathway area based on the information of ruts and dead grass, it is difficult for robots to determine the direction of travel. However, in the case of pathways covered with plants, it is difficult to estimate the pathway area because the pathway boundary is ambiguous. Therefore, we propose a method to estimate the path directly on the image. To confirm the usefulness of our method, we build an estimator and conduct a navigation experiment in an outdoor pathway. We create training data, and train it on the constructed Convolutional Neural Network (CNN) architecture. After training, the estimation accuracy was more than 90%, and we were able to run the navigation experiment in Sakaki Park in our university.

The purpose of this study is to model the awareness of a pedestrian and to develop a pedestrian simulator based on the Social Force Model considering the model. In order to model the awareness, we used three factors: the viewing angle, the distance, and the update period. The viewing angle and the update period of the awareness are modeled so that they can be simulated with arbitrary values. The distance is modeled to have a probabilistic awareness depending on the distance. To evaluate the pedestrian simulator with the proposed awareness model, we compared the walking trajectory of the simulator with that of the real world, and evaluated the model based on the error of the comparison.

In this paper, we present an improved motion planning algorithm for mobile robots operating in indoor environments. Collision free robot motion planning in obstacle rich environment is crucial for the safety of the robot and for other agents or people in the vicinity. There are a plenty of motion planning algorithms that have been proposed in the past. In this particular work, we present a local planning scheme that allos the robot to overcome static and dynamic obstacles in the path with high success rate. Our proposed reactive planning algorithm uses a potential field for generating a collision-free paths for the robot. The proposed algorithm can overcome some of the challenges of a deadlock situation of traditional planning and guarantee successful navigation for robots.

Due to the social and technological background, the number of robots that can coexist with humans and perform autonomous navigation is increasing. In order to facilitate the implementation of robots in society, it is necessary to increase the comfort level of autonomous navigation. However, conventional autonomous navigation without human recognition is said to contain elements that are uncomfortable for humans. These unpleasant elements are low legibility and intrusiveness to proxemics. In order to eliminate these uncomfortable elements, we propose an autonomous navigation method with an attention state estimation function based on human action recognition and a human proxemics recognition function. In addition, we have implemented the method and confirmed its feasibility by running the system experimentally.

The 2-DOF spherical parallel link mechanism (2-DOF SPM) has been applied to the direction operation of cameras because the error due to link deformation and the torque applied to the actuators are small. However, the range of motion is narrow because it is easy to transfer into a singular posture and parts interference is likely to occur. This study proposes a slider-type 2-DOF SPM with a wide range of motion. In this mechanism, the input rotation axes are arranged coaxially, and the end effector has sliders at both ends. We also applied a design method eliminating the component interference factors to the other slider-type parallel link mechanism, diversified the configuration, and expanded the range of motion.

To measure cutting force of scissors without the cutting position influence, we developed a pair of scissors which employs a transmission using a curved slider. This paper proposes the design of the curved slider that reduces the cutting force error depending on the cutting position. The cutting force error is caused by the difference in shape of the scissors and the curved slider. Furthermore, the attachment angle between scissors blade and curved slider does not affect the error depending on the cutting position. We confirmed the scissor blade shape using a document scanner, and found that the experimental results of the cutting force error is agreement with the analysis results.

Since the parallel mechanism does not add the inertia of the motor to the end effector, it is easier to reduce the weight and speed up of the end effector compared to the serial robot. A method has been proposed in which the disturbance torque around the motor rotation axis is compensated by the disturbance observer to improve the target value tracking characteristic and the disturbance suppression characteristic. However, if the stiffness of links is low, the control system tends to become unstable. Therefore, in this paper, a link bending moment is fed back to a position controller to suppress vibration of the end effector. Validity of the proposed controller is verified by experiments.

Hybrid Stepper Motors are used in a closed-loop form, which utilizes an incremental encoder in order to keep track of missed steps. In this case, the encoder-motor missalignment is neglected most of the time. Previous research conducted on this topic are based on a single measured values. This makes it difficult to understand the step distribution over every encoder ticks. The Naive Bayes method is introduced to model the rotor-encoder mapping, which is used to correct the encoder-ticks to motor-step table. The rotor position used for modelling is measured with different current values. The effectiveness of the proposed method is shown through the reduction of absolute error between the positive and negative velocities.

In this paper, kinematics and dynamics of a four-dof differential belt and screw driven parallel robots are analyzed by the multi-body systems using a symbolical formula manipulation. Mechanical characteristics of the idle pulleys and slides, which are difficult to consider by the conventional input/output (from the motor to the end effector) analysis, can be included in the generalized mass and viscosity matrices of the robot by our study. The proposed method was installed by Mathematica then applied to derive the kinematics and dynamics of the robot. The derived results suggest that the robot may be able to perform the iso-inertia and low-interference design.

In this study, the optimal control of a small ship transfer system is performed by considering the ship’s sway that exceeds the range of motion and by incorporating an evaluation index that includes the amount of change as well as keeping the output plate horizontal.

A motion table system in 3D space is proposed and its kinematics and inverse kinematics are calculated. The table is suspended by four wires and it can be positioned in an arbitrary point in its working space. Since the number of wires suspending the table are not enough to achieve the desired orientation of the table, a counterweight is installed those position is controlled by a positioning system on the table. The kinematics and inverse kinematics of the system including the counterweight control system are calculated and its characteristics are analyzed.

Mobile robots with both high trampling ability and low COT are required for hazard removal and disaster rescue in unstructured environments. In this paper, we present a Spider-Man inspired mobile robot that can brachiate using wires to overcome obstacles. The proposed mechanism has two wire injection mechanisms and winding mechanisms. By ejecting the two wires toward a ceiling and rewinding them, the robot can control its position and angle for brachiation. Then, by detaching one of the two wires, the robot starts brachiation and lands on a remote point. We build a prototype robot and experimentally verify that it can brachiate along the target trajectory.

In agriculture work, weeding work in paddy field has been physical burden and taken a lot of working time. As a solution of this problem, stopping the growth of weed by stirring the soil is affective. An example is "Duck farming" that ducks stir the mad by swimming in paddy fields. In this paper, we propose new robot that imitate snakes to stir the soil. A body segment of the snake-like robot is composed of three bending modules , continuous joint mechanism, layered gear and two wire driving mechanisms, we approximate the serpenoid curve by curves without using straight lines.