Keywords: Soft Robotics, Mechanism Design
Abstract: Fiber reinforcement is a well-studied approach for fabricating soft actuators. In this approach, an inextensible fiber is employed to constrain the deformation of a membrane in the fiber axial direction. In this way, the desired deformation can be achieved. In conventional finite element modeling, usually, the fiber and the membrane are modeled separately and then combined into a single model by imposing constraints. However, this is a time-consuming process. We propose a new approach to implicitly model an inextensible fiber by imposing geometric constraints, namely, the constant edge-length constraints. In our approach, the fiber model is integrated into the membrane model as mesh edges, and the two models are combined into one. As a result, the modeling process is simplified. The simulation results show that by employing the proposed model, the deformation behavior of fiber-reinforced soft structures can be simulated.

Keywords: Welfare systems, Systems for Service/Assistive Applications, Human-Robot/System Interaction
Abstract: Maintaining a healthy parent-child relationship in a family is paramount and has a significant effect on the mental well-being and development of children. One way for improving parent-child communication and relationship is the Parent-Child Interaction Therapy (PCIT), which is a well established therapy method for preschool children and their parents, and has proven to be highly successful. Employing PCIT requires well trained and certified therapist personnel, and a specially constructed therapy room. These requirements make it hard to adopt and practice PCIT for families who could profit from the therapy. In order to help to spread this successful therapy to more children and their parents, we propose a therapy robot system to aid those already taking part in the therapy. This way they can perform the therapy interaction tasks in their homes more effectively and without direct therapist supervision. The proposed system listens to the conversation between the child and parent, and based on the Dyadic Parent-Child Interaction Coding System (DPICS), gives appropriate reactions, so the parent receives feedback whether the interaction is going well or not, thus potentially improving the interaction between parent and child. This system uses Natural Language Processing and Understanding for extracting the necessary information from speech to drive a fuzzy automaton-based behavior model optionally attached to interfaces of social robots. We believe that in the near future, such robot system can improve the efficiency of the independent home users and can also reduce the burden on the therapist personnel. Furthermore, the proposed therapy assisting social robot system can be adopted outside parent-child interaction in the future, e.g. in elderly care for improving caregiver-patient interaction.

Keywords: Mechanism Design, Mechatronics Systems
Abstract: Resin 3D printing is a promising modeling method for Agile development, however its material strength is lower than that of metal parts. We hypothesized that embedding a thin-walled metal pipe in a 3D printed beam would increase the strength of the part while keeping it lightweight. This paper aims to verify the effectiveness of this method. To test our hypothesis, specimens of potassium titanate fiber-reinforced material POTICON and ABS were fabricated, and conducted three-point bending tests. As a result, POTICON specimens embedded with stainless steel pipes showed a strength 0.7 to 1.4 times that of the solid specimens, while ABS specimens showed a strength 1.1 to 2.0 times greater than the solid specimens. These values were small compared to our flexural rigidity calculations, and we concluded that embedding a pipe did not significantly increase the strength.

Keywords: Mechanism Design, Mechatronics Systems, Hardware Platform
Abstract: Lightweight industrial robots are increasingly required to reduce power consumption. In this research, we focus on lightweight resin materials such as 3D printed parts, and our ultimate goal is to develop practical industrial robots using these materials as structural parts. Unlike conventional light metal materials such as high-strength aluminum, resin materials have creep deformation properties even at room temperature, which deteriorates the end-point position accuracy in practical use over a long period. In this paper, we fabricated an experiment apparatus that can measure creep deformation by applying 40 kgf of load at the end of a beam specimen. A laser displacement meter accurately measured the creep deformation for approximately two months (1600 hours). For comparison, we measured the creep deformation of six specimens of five different lightweight resin materials and conventional aluminum alloys (A5052). The results showed that 3D printed materials had large creep deformation, whereas A5052 and FELCARBO showed almost no creep deformation.

Keywords: Mechanism Design, Factory Automation, Mechatronics Systems
Abstract: To reduce the weight of a mechanical arm, 3D printing is gradually being applied to the field of robotics. A lightweight robotic arm indirectly reduces the torque required of the motor and reduces the overall cost and energy consumption of the robot. However, with 3D printing, there is no standard for choosing the filling density. A low filling density means fast printing, cost saving, and light weight. A high filling density results in superior bending strength and more perfect appearance. Compared with high-rigidity metals, resin is more easily affected by vibrations. To apply 3D printing technology to industrial robotic arms and other applications that require motion precision in future, we start with the influence of the filling density on the vibration characteristics, take the settling time required for the manipulator to return to rest after vibration as the criterion. The optimal filling density and the applicable occasions of different filling densities are discussed. According to the experimental results, printing with low filling density is preferred in applications that require less load or impact on the robotic arm such as chip manufacturing. For robotic arms that require high-load operations or high impact on the arm, high-fill-density printing is preferred.

Keywords: Soft Robotics
Abstract: Flexible and lightweight soft actuators have attracted significant attention in recent years. One such soft actuator is the Ionic Polymer Metal Composite (IPMC) actuator, which can operate in water, produce low noise, and be driven at a low voltage. This actuator has a thin-film structure with an ion-exchange membrane coating, and performs a bending motion when a voltage is applied to its electrodes on both sides. Herein, we apply this actuator to a tetherless IPMC robot with a high degree of freedom by eliminating wiring from the power supply and simplifying the components. Focusing on the power source, we propose a method for depositing and adhering electrodes using sputtering equipment to integrate batteries with IPMC actuators. A magnesium battery that can function as an electrolyte in water is used, which provides the operating environment of the actuator. In this study, electrode deposition is performed using sputtering equipment, and the sputtering conditions for magnesium were examined. In addition, three prototypes were fabricated and compared with metal plates, and experiments were conducted to survey the actuator behavior with the battery electrode under the deposition.

Keywords: Human-Robot/System Interaction, Systems for Service/Assistive Applications, Human-Robot Cooperation/Collaboration
Abstract: As social robots play an increasingly active role in nursing care and medical fields, opportunities for robots to communicate with people of all ages are expected to increase. It is important for robots to persuade people to act in order to encourage them to achieve a specific goal. We believe that it is necessary to identify a strategy for robot behavior that can encourage people to take action to finally achieve their goals, even if they refuse to be persuaded by the robot. Therefore, we propose a strategy that incorporates not only robot behaviors that people feel comfortable with, but also dares to incorporate uncomfortable behaviors at effective times. In this strategy, we set up two types of robot behaviors: rewarding behavior and punitive behavior. Reward behaviors are favorable, expresses positive empathy, and is willing to accept instructions and cooperatively perform actions. Punitive behaviors are not favorable, expresses negative empathy, and makes people feel uncomfortable, such that they do not want to accept instructions and cooperatively perform actions. In this paper, we conducted a study on adult and older adult subjects to investigate the impression of the robot's gaze behavior, gestures and voice, which were designed as rewarding and punitive behaviors.

Keywords: Machine Learning, Human-Robot/System Interaction, Human Interface
Abstract: As a hands-free device for personal mobility vehicles, PMVs, we have thus far presented a brain-machine-interface, BMI. The BMI, however, occasionally classified electroencephalography, EEG, signals into wrong control commands. As a result, a PMV exhibited inappropriate behavior and was in danger of collisions with curbs. In order for the PMV to avoid the collisions, we propose a fail-safe controller. For control commands from the BMI, the fail-safe controller determines the final control commands in consideration of surroundings. For this purpose, a depth map that estimates the depth values of an input image is simultaneously generated. The fail-safe controller is based on deep neural networks, such as convolutional autoencoder, CAE, and fully-connected layers. The CAE is used to derive depth maps, and final control commands are determined from feature maps extracted by the CAE through the fully-connected layers. For these multiple outputs, i.e., depth maps and final control commands, multi-task learning is applied to the fail-safe controller. The CAE is trained through deep convolutional generative adversarial network, DCGAN. The fully-connected layers are trained through supervised learning. Furthermore, we introduce undesired control commands that cause collisions into a dataset, especially when the PMV is close to curbs. Finally, we show the effectiveness of the proposed fail-safe controller for collision avoidance.

Keywords: Machine Learning, Vision Systems, Systems for Field Applications
Abstract: Plant disease diagnosis is a very important task for maintaining agricultural productivity. Convolutional neural networks (CNNs) are popular tools in the field of image recognition and have been widely used in plant disease diagnosis. However, they are prone to small data set overfitting and their inability to quantify uncertainty in their predictions can lead to overconfidence and errors. To address these issues which hinder the construction of practical plant disease diagnostic models we propose the use of Bayesian convolutional neural networks (BCNNs). BCNNs are robust against over-fitting regardless of the size of the data set and the probability distribution of the models' weight parameters can be estimated to obtain predictive distributions whose standard deviation expresses their predictions' confidence. In this work, we develop a BCNN for plant disease diagnosis and perform comparative experiments between conventional CNN and our BCNN for plant disease diagnosis. Our results show that our BCNN model is more robust to over-fitting than the CNN model on small data sets and that the confidence in its predictions can be when diagnosing plant diseases.

Keywords: Machine Learning, Plant Engineering, Vision Systems
Abstract: In this study, we propose a method for detecting changes on the outer surface of pipes using inspection videos captured by an inspection robot. It is critical to detect anomalies on the outer surface of pipes during patrol inspections. Anomalies are defined as deviations from the normal state and should be detected as areas that have changed from the normal state. Therefore, for appropriate maintenance of the plants, it is crucial to perform change detection by comparing videos that capture the past normal state with those capturing the current state. The problem with detecting changes from videos is deciding which frame to compare. We therefore propose sequential filtering to determine image pairs based on the position of the images and their similarity. We then apply a deep learning method to perform change detection. An indoor simulated plant environment has been constructed to test the efficacy of the proposed method. Experiments and evaluation results showed that the proposed method outperformed an autoencoder. The proposed method also achieved an F1 score of 0.880 for change detection in the inspection videos by introducing sequential filtering, which prevented mismatching of image pairs and reduced computational costs.

Keywords: Systems for Field Applications, Environment Monitoring and Management, Machine Learning
Abstract: With the gradual decrease in fisheries resources and the increasing demand for fishery production, the production of capture fisheries is saturated. On the other hand, aquaculture production is increasing rapidly and now accounts for almost half of all fishery production. In aquaculture, monitoring aqua cages and collecting basic information, such as the number of fish in the cages and their sizes, is one of the most crucial tasks. However, automation of these basic tasks is insufficient, and there is an expectation for the implementation of ICT (Information and Communication Technology) in the fisheries industry. To move towards smart fishery farms, efficient fisheries resource management is required.　This paper introduces a computer vision-based fish counting system that utilizes image enhancement and machine learning algorithms to achieve automated fish counting in aqua cages. Experimental results demonstrate that the proposed method can detect fish and identify the same fish in images captured by underwater cameras set at different depths. The proposed method can contribute to fisheries management and ecological conservation.

Keywords: Systems for Field Applications, Environment Monitoring and Management, Intelligent and Flexible Manufacturing
Abstract: In the agricultural field, many issues arose such as the decline of productivity due to population aging and the lack of new generations to replace the previous ones. Information and communication technology (ICT), the Internet of Things (IoT), and robotics are expected to provide solutions. Despite the enormous promise of these technologies, they are expensive to implement, require ongoing technical assistance, and have little affinity for the immediate demands of farmers. For example, when a new breed is improved, it is necessary to ask the developing company to outsource data analysis to identify defective new individuals. Therefore, the main objective of this study is to realize an economically viable and easily implementable smart agricultural system. For this purpose, interviews were conducted with actual farmers, and it became clear that even in the fruit sorting process, which is becoming increasingly automated, the burden on farmers with knowledge and skills is significant. In this study, we focus on mandarin oranges, which have the highest yield among Japanese oranges, to automate the identification of defective fruit, i. e. damaged or affected by a disease, which has been done by the human eye. In this study, an automatic fruit sorting system based on a conveyor belt mechanism was designed. The system performs fruit sorting using a CNN based on individual features derived from images, but since the number of images of defective individuals is far fewer than the number of images of good individuals, the system was supplemented by a Cycle-GAN. This integrated approach aims to alleviate labor intensity and enhance operational efficiency, thereby contributing to the promotion of smart agriculture.

Keywords: Machine Learning, Vision Systems, Automation Systems
Abstract: In this study, we proposed a training method using joint points obtained from physical simulations and real image data for the action recognition of excavators. The proposed method classifies the action recognition of excavators into position detection, skeleton detection, and action recognition models. The first two models are trained using the real image data, whereas the action recognition model is trained using the joint point data obtained from the physical simulation. For the action recognition model, we proposed a data augmentation method based on the features of the actions of the excavator. Experimental results indicate that the proposed method can achieve better accuracy than the conventional method that uses real video data, though the proposed method does not use any real video data for training.

Keywords: Vision Systems, Automation Systems, Environment Monitoring and Management
Abstract: Image segmentation is one of the critical tasks in UAV-based infrastructure monitoring. The evaluation of a segmentation is conducted by comparing the ground truth annotated by a human expert to a resulting image by an algorithm. In the absence of the ground truth, it is not easy to evaluate the performance of the deployed models. Hence come empirical goodness metrics, where segmented images are compared with the original one using statistical measures. In this paper, we analyze the correlation between the F1 score and a well-known ground-truth-independent metric, the region contrast C, from the segmentation results of 9 crack detection models on the SYDCrack dataset. Experimental results have confirmed the alignment between C and F1 regarding model ranking. However, the correlation analysis does not show a linear dependence between the two metrics on observations of all models. In fact, the region number of segmentation is highly correlated to both evaluation metrics. This might shed some light on developing an empirical goodness metric for vision-based crack detection.

Keywords: Sensor Networks, Machine Learning, Decision Making Systems
Abstract: The growing number of applications for unmanned aerial vehicles operating in close proximity to humans calls for strict safety requirements. To ensure reliability and safety, fast and effective diagnosis of possible drone faults is needed. In this paper, a study on the detection of damage to the propellers of a multirotor using inertial sensors was conducted. Measurement data from the publicly available PADRE repository were used. Frequency features and raw measurement data applicability were tested. Two different approaches to data processing during fault detection and classification were tested. The first is to analyze the signals after collecting the entire set of data needed for treatment by the neural network. The second approach processes the data after every single acquisition of sensor measurements. The performance results and processing time of each solution are recorded for analysis. By effectively selecting parameters of the proposed approaches, the processing time and accuracy of UAV fault classification can be significantly improved, as verified over different classes of propeller defects. Early fault detection is essential for safe operations of multirotor drones.

Keywords: Integration Platform, Hardware Platform, Mechatronics Systems
Abstract: More and more steel bridge collapse accidents occur worldwide due to broken steel structures, causing significant loss of life and property to mankind. This has spurred much research into robots that can climb steel surfaces and carry smart sensors with the desire to assist inspectors in inspecting steel defects. However, current non-destructive evaluation (NDE) sensors such as eddy-current and giant magnetoresistance (GMR) are not able to be integrated with the robotic platforms due to their large size, mass, and limited ability to interface. More importantly, these NDE sensors may fail to detect hidden and underlying cracks. Therefore, in this paper, we present a novel and compact analog magnetic sensor system to detect different types of cracks in steel. Validations are carried out on a steel test plate 600mm long, 140mm wide, and 6mm thick, and precision machined by CNCs to produce various man-made cracks including penetrating cracks, surface cracks, internal/hidden cracks, and underlying cracks. These cracks have a width varying from 0.6mm to 1.2mm with different depth levels. Combined with applying the Kalman filter to the noise reduction system, the results obtained are accurate, and the response speed is fast. The sensor is small enough and has firmware/software with an interface to make it possible to be integrated with a small drone or climbing robot to perform an in-depth inspection of the steel bridge. A demonstration of the system can be seen in this video: https://youtu.be/OrdpLdnF1Yk

Keywords: Path Planning for Multiple Mobile Robots or Agents, Multi-Robot Systems
Abstract: This paper presents a new swarm intelligence-based approach to deal with the cooperative path planning problem of unmanned aerial vehicles (UAVs), which is essential for the automatic inspection of infrastructure. The approach uses a 3D model of the structure to generate viewpoints for the UAVs. The calculation of the viewpoints considers the constraints related to the UAV formation model, camera parameters, and requirements for data post-processing. The viewpoints are then used as input to formulate the path planning as an extended traveling salesman problem and the definition of a new cost function. Ant colony optimization is finally used to solve the problem to yield optimal inspection paths. Experiments with 3D models of real structures have been conducted to evaluate the performance of the proposed approach. The results show that our system is not only capable of generating feasible inspection paths for UAVs but also reducing the path length by 29.47% for complex structures when compared with another heuristic approach. The source code of the algorithm can be found at https://github.com/duynamrcv/aco_3d_ipp.

Keywords: Software Platform, Integration Platform, Vision Systems
Abstract: In this paper, we proposed a novel multi-camera auto-adjustment framework for infrastructure inspections, specifically 3D point cloud mapping via lidar-camera fusion. Our method uses a single controller for consensus target brightness and adaptive lighting. In addition, a vignette correction filter was created to apply to images outputted from cameras to avoid dark strips on the point cloud data. This approach is designed to be a versatile, real-time friendly, and platform agnostic software stack.

Keywords: Machine Learning, Vision Systems, Surveillance Systems
Abstract: This work presents a neural network model capable of recognizing small and tiny objects in thermal images collected by unmanned aerial vehicles. Our model consists of three parts, the backbone, the neck, and the prediction head. The backbone is developed based on the structure of YOLOv5 combined with the use of a transformer encoder at the end. The neck includes a BI-FPN block combined with the use of a sliding window and a transformer to increase the information fed into the prediction head. The prediction head carries out the detection by evaluating feature maps with the Sigmoid function. The use of transformers with attention and sliding windows increases recognition accuracy while keeping the model at a reasonable number of parameters and computation requirements for embedded systems. Experiments conducted on public dataset VEDAI and our collected datasets show that our model has a higher accuracy than state-of-the-art methods such as ResNet, Faster RCNN, ComNet, ViT, YOLOv5, SMPNet, and DPNetV3. Experiments on the embedded computer Jetson AGX show that our model achieves a real-time computation speed with a stability rate of over 90%.

Keywords: Mechatronics Systems, Systems for Service/Assistive Applications, Human Factors and Human-in-the-Loop
Abstract: In this paper, the authors present an intelligent lure which can avoids rooting by applying a disturbance observer (DOB). This environment-friendly lure has a propeller motor for backward motion. We make a better use of this motor to classify the lure’s motions into sinking, being drawing, and hitting something according to a time-series of DOB estimates. Its prototype has been implemented and its ground experiments have been conducted. Experimental results prove we have extrapolated the lure’s motions successfully.

Keywords: Medical Systems, Human Interface, Machine Learning
Abstract: Artificial Intelligence (AI) is used for various fields including the medical field. A lot of data are necessary to improve the accuracy. However, it is difficult to gather the medical data. This paper describes the AI system that improves the accuracy with every usage automatically by the online improved techniques applied to the diagnosis of rheumatoid arthritis (RA). And its effectiveness is predicted. The result of the running using two PCs and one server and the realization of the system are shown. As a result, we confirmed that the serve-client system that is to update the model automatically using uploaded data set made by each client is constructed.

Keywords: Human Factors and Human-in-the-Loop, Human Interface, Vision Systems
Abstract: This paper proposes a three-dimensional aerial projection device, All-Round Accommodator, with rich reality using Peppers Ghost and variable focal length lens. In order to achieve improvement of rich reality of aerial projection, our proposed device incorporates two factors of motion parallax and accommodation. The former achieves all-round view by using the property that the pepper's ghost can be observed from many directions. We can perceive three-dimensional effect when projection is moving. The latter is achieved by combining a volumetric display and a variable-focal lens. An advantage of accommodation is to realize three-dimensional reality even when projection is not moving, i.e., still. In order to prove our system is feasible, a prototype zero for demonstration has been designed, implemented and experimented if we can feel three-dimensional effect from aerial projection. Experimental results have shown our principle performed properly.

Keywords: Human Factors and Human-in-the-Loop, Modeling and Simulating Humans, System Simulation
Abstract: Robot simulators are commonly employed in the study of autonomous mobile robots operating in human-robot coexistence environments. In these simulators, it is essential to place moving pedestrians as dynamic obstacles, which requires the application of various crowd models developed in the field of crowd simulation. However, existing crowd models often assume that all agents, including both pedestrians and robots, follow the same navigation algorithms. The validity of using such models for robot simulators remains unclear. This paper aims to explore the impact of robots with different navigation methods than pedestrians and the environment in which they are placed on the behavior and performance of crowd models in a robot navigation simulator. The experimental environment is constructed within a mobile robot simulator in which multiple agents, including robots, are deployed and crowd simulations are performed. These simulations are performed by moving the agents through multiple crowd models. In addition, the impact of the presence of robots and environmental factors on the performance of the crowd models is evaluated. The results indicate that the performance of the crowd models is highly dependent on the environment in which the robot operates, and that it is necessary to select the appropriate model for a given environment.

Keywords: Human-Robot/System Interaction, System Simulation, Systems for Service/Assistive Applications
Abstract: This study proposes a shopping cart system with constant steerability regardless of the loading weight and position. Conventional carts are often affected by varying loads, necessitating corrective steering or exertion of excessive operating force, leading to decreased comfort and safety. To address these challenges, we propose a steering assist system that integrates passive robotics concepts. In this paper, we present a steering control method that combines translational and rotational movements. Furthermore, we validate this proposed method by utilizing a steering simulator that we have developed.

Keywords: Rehabilitation Systems, Systems for Service/Assistive Applications
Abstract: In this study, in order to perform standing support that takes into account differences in the muscle strength between individuals, it is necessary to change the support movement according to the muscle strength status of the individual. Therefore, it is necessary to measure the muscle strength of individuals, and most previous studies measured lower limb muscle strength by using electromyographs. However, because the person using the electromyograph needs expertise in muscle strength measurement, it cannot be easily used by ordinary caregivers in the care setting. In addition, measurement is difficult when using the electromyograph because it requires the attachment of many electrodes to the subject's body. At the same time, many previous studies have been conducted on ideal standing movements. The purpose of this study is to propose a stability margin range according to the subject's physical fitness. We proposed a simple method for estimating lower limb muscle strength without using an electromyograph and imposing no burden on the subject and the measurer. Based on the estimation results, we use the body center of gravity and the amount of muscle exertion as indicators and propose a stability margin range according to the subject's physical fitness

Keywords: Factory Automation, Mechatronics Systems, Plant Engineering
Abstract: The manufacturing industry is increasingly focused on sustainability, requiring greater product transparency to meet sustainability targets. Digital Product Passports (DPPs) have emerged as a potential solution to improve recyclability and inform consumer purchasing decisions. However, there is a lack of methodology and digital representation for building DPPs, which hinders data linkage and sharing. In this paper, we propose an approach based on standardised data models and event sourcing to aggregate production data, using energy consumption as an example. We present a concept that captures and stores process information using event sourcing in Asset Administration Shells (AAS), and evaluate its feasibility through a manufacturing demonstrator. The results show that event-based aggregation facilitates the creation of DPPs, allowing different actors in the value chain to participate in an interoperable way. By leveraging a standardised and interoperable common data space, this approach improves product transparency, promotes recyclability and enables informed consumer choice in line with sustainability goals.

Keywords: Mechanism Design
Abstract: The double-row active omni wheel (DAOW) has been proposed to achieve omnidirectional mobility with low vibration and a simple structure compared to the conventional design. According to a kinetic analysis, the DAOW was found to have a stabilizing effect on single-track vehicles in the roll direction, but this phenomenon has not yet been confirmed in the real world. With this in mind, this paper aims to experimentally verify the posture-stabilizing effect of the DAOW. We first analyze the results of a previous study and discuss the conditions for establishing the stabilizing effect. Then, we design a modified prototype vehicle and conduct an experiment. The result indicates that the prototype vehicle maintains its balance for a certain period and exhibits the behavior of the stabilizing effect.

Keywords: Mechanism Design, Systems for Field Applications, Hardware Platform
Abstract: This paper proposes a prototype design of tracked in-pipe inspection robot that possesses two functions: translating along the pipeline axis and rotating within the pipeline, with the ability to switch between these two functions. This robot requires only a minimal complement of ultrasonic sensors, enabling it to accomplish a 360-degree wall thickness inspection by means of rotation. This approach significantly contributes to cost savings in the inspection process. The design of this robot consists of the track, rotation, and switching modules. The track module is a mechanism moving forward or backward for this robot, The rotation module is used for the motion of the robot turning in the pipe. The switching module allows for seamless transition between the track and rotation modes. The switching module is driven by a single motor and incorporates a locking mechanism to ensure that at least one pair of modules support the robot against the pipe wall during the mode transition, thereby maintaining the robot's posture. To validate the design, motion simulations were conducted using SolidWorks. Then, we performed movement tests on the track, rotation, and switching modules in a real pipe. The test results confirm the feasibility of the prototype design.

Keywords: Mechanism Design, Mechatronics Systems, Systems for Field Applications
Abstract: Overcoming labor shortages is a major challenge in global agriculture. One approach to address this is to develop autonomous weeding robots that can remove weeds from crops without human intervention. Our group has been developing an autonomous weeding robot that can pull weeds with improved weeding speed. This study proposes a new weeding device for autonomous robots that can perform weeding operations quickly. The weeding device has tweezers that pinch and pull weeds onto a rotating cylinder. Although the proposed weeding device can pull weeds, the success rate is not high, and weeds are often torn. Moreover, challenges, such as pulling out the weeds incorrectly and getting weeds tangled in the tweezers, remain.

Keywords: Mechanism Design
Abstract: This paper proposes a cleaning machine using a planetary gear mechanism to remove accumulated grease within kitchen ducts of diverse geometries, encompassing circular and square configurations. The contributions of this paper encompass describing the brush and motor determination methods, detailing the development of the actual machine, and presenting the outcome of the cleaning experiment. The purpose of this study is to remove accumulated grease within ducts via a single robot, thereby preventing duct conflagrations. While duct cleaning is typically done manually, small, hard-to-access ducts have often been uncleaned. In this study, we developed a cleaning machine using a planetary gear mechanism capable of scrubbing off accumulated grease within small square ducts through a singular brushing operation. In the cleaning experiment, the proposed machine successfully removed over 30% of the accumulated grease within a small, difficult-to-access duct. Therefore, we expect that the proposed machine will make a significant contribution to the prevention of duct fires by scrubbing off grease from ducts that are typically difficult for people to access and have not been cleaned in the past.

Keywords: Mechanism Design, Biomimetics, Mechatronics Systems
Abstract: A snake-like robot can traverse rough terrain where conventional robots cannot. Many previous studies focused mainly on locomotion of snakes. However, snake locomotion still has the problem of high power consumption due to the drive of many actuators. To solve this problem, a method is proposed in which a snake-like robot is transformed into a wheel shape and the center of gravity is moved and the robot rolled by the gravitational force. This motion method was experimentally shown to be more efficient than the conventional locomotion of a snake-like robot. On the other hand, a simulation using a model of a snake-like robot transforming into a wheel shape did not fit the experimental results. Therefore, a modified model is proposed by considering the nonlinear term of the dynamics and investigating an interpolation method of the center of gravity trajectory.

Keywords: Factory Automation, Soft Robotics, Automation Systems
Abstract: The aim of this study is to improve the cycle time of transport operation by a robotic manipulator equipped with a soft robotic hand. This paper proposes a novel optimization strategy for the manipulator motion that considers the vibration induced in soft robotic hands. Using Bayesian optimization, the manipulator’ motion parameters are optimized through a small number of actual experiments. In the experiments, the cycle time is measured using only a 30-fps camera, resulting in the low-cost parameter tuning system. The proposed method was experimentally evaluated by comparing the result obtained by the proposed method and the grid-search-based tuning method. The obtained cycle times were close to each other, and the proposed method achieved the optimization with 15% of the number of searches in the grid search.

Keywords: Machine Learning, Vision Systems
Abstract: Texting while walking is a common behavior exhibited by pedestrians. While several studies explored the detection of texting while walking, the influence of occlusions were neglected. In this paper, we propose an image-based method which utilizes a pre-trained Variational Autoencoder. The proposed method takes sequence of 2D coordinates of upper body key points of the pedestrians as input, encodes the data into a 2D latent space, and uses the encoded data to distinguish text walkers from normal pedestrians. The proposed architecture enables the model to extract meaningful features from occluded data. Results of ablation test and comparison with a previous method revealed that the proposed architecture is successful in identifying text walkers even under heavy occlusion, outperforming a previously proposed method.

Keywords: Systems for Field Applications, Automation Systems, Mechatronics Systems
Abstract: This paper describes an agricultural robot for strawberry fruit harvesting and truss pruning. Although various agricultural robots were studied, most of them had single- function. The target fields for these are also a well-maintained environment in which they are relatively easy to introduce the agricultural robots, such as a plant factory. The field in which the developed robot in this study operates is small-scale greenhouse, and this robot is a multifunctional rather than a single-function agricultural robot. This robot consists of the following: a crawler-type moving mechanism, a manipulator with 6 degrees of freedom, an end-effector capable of cutting and gripping, and an RGBD camera. In addition, to realize the functions of harvesting and pruning, a method of recognizing strawberry plants and detecting cutting points for both harvesting and pruning is implemented. Harvesting and pruning experiments was conducted at a strawberry greenhouse. This paper reports the results. For the results of the harvesting experiment, the harvesting time per fruit was approximately 20 seconds, and the success rate was 80%. For the result of the pruning experiment, the pruning time per truss was approximately 23 seconds, and the success rate was 80%. These experimental results demonstrate the effectiveness of the multifunctional agricultural robots.

Keywords: Multi-Robot Systems, Energy and Environment-Aware Automation, System Simulation
Abstract: In this study, we propose a methodology for automating earthmoving tasks using autonomous excavators and dump trucks, with an aim to enhance adaptability in response to environmental changes. Our methodology involves a general architecture that fosters self-organized swarm behavior among the robots. Furthermore, we present an illustrative algorithm that evaluates the performance of a team of robots and adjusts the team composition based on this performance evaluation. This algorithm was implemented in a simulation test in a dynamic environment. The results demonstrate that our methodology enables the coordination of excavators and dump trucks in environmental changes.

Keywords: Systems for Field Applications, Machine Learning, Autonomous Vehicle Navigation
Abstract: For mobile navigation of mobile robot in uneven outdoor terrain environments, recognition of terrain traversability is still a challenging problem. One promising approach to this problem is self-supervised learning, which relies only on the robot sensor information both for exteroceptive sensing as input and proprioceptive sensing as output. In order to reduce the cost to run over various terrains by the robot, efficient sampling of the training dataset for the terrain traversability prediction is of great importance. Toward this end, this paper presents an evaluation of confidence-aware prediction framework of terrain traversing risk. A CNN model with Monte Carlo dropout was implemented for the predictor. In the experiment, it was confirmed that the traversing cost can be predicted well for various uneven natural terrains. The confidence-aware approach was evaluated in the experiment and some future directions to utilize the confidence were discussed.

Keywords: Multi-Modal Perception, Machine Learning, Sensor Fusion
Abstract: We propose an ego-noise prediction method applicable to Unmanned Aerial Vehicles (UAVs). Sensing sound using UAV has a significantly reduced Signal-to-Noise Ratio due to the strong ego-noise from the propellers and motors. Template-based ego-noise prediction for robots with pre-set movements, such as humanoid robots, has previously been studied. However, that approach is not applicable to UAVs which continuously adjust the propeller’s rotation speed based on the situational requirements. Here, we propose a novel framework for ego-noise prediction by focusing on propellers, which are the main ego-noise source, and translating the modality of propeller vibrations into the modality of ego-noise sound. Assuming a strong relation between the vibration frequency of a propeller and the ego-noise it generates, we propose using a Recurrent Neural Network to predict the ego-noise spectrogram, using vibration frequency as a feature. The prediction accuracy of ego-noise spectrograms was at a level that suggests the potential for effective ego-noise reduction, and the results are promising for applications in UAV hammering inspection.

Keywords: Systems for Field Applications, Mechanism Design, Mechatronics Systems
Abstract: This paper presents the experimental results of energy consumption of a wheel/track reconfigurable mobile robot which has been developed for maintenance of vineyard. According to the Bekker's soil model, the wheel rolling resistance depends on the sinkage to the ground, so that the mobile mechanism consists of 2 rear wheels and 2 front tracks which is designed to lift up and down and control the contact area to the ground and to switch the modes based on the soil conditions to minimize energy consumption. The wheel mode is supposed to be operated for the firm soil where the front tracks are lifted up to decrease the contact area, and the track mode is for the soft soil where the front tracks push down the ground and increase the contact area and decrease sinkage. In the experiment, we compared the energy consumption between the wheel mode and the track mode in firm, soft, and wet soil conditions.

Keywords: Autonomous Vehicle Navigation, Hardware Platform, Software Platform
Abstract: This document presents the implementation of a fixed-wing UAV for online control using advanced algorithms techniques. First, the aerodynamic and kinematic models are presented. Moreover, hardware for autonomous and controlled flying capabilities is presented, considering parallel programming capabilities, integration with Ardupilot firmware and communication with different protocols. An example of a computationally intensive control algorithm is showcased to highlight the advantages and reliability of the proposed approach. Finally, simulations and experiments are performed to validate the proposal.

Keywords: Motion and Path Planning, Autonomous Vehicle Navigation, Automation Systems
Abstract: This paper introduces a Grey Wolf Optimization(GWO) technique for addressing path planning challenges of unmanned aerial vehicles when inspecting an environment with multiple obstacles. By formulating a comprehensive cost function, the problem is cast as an optimization task that accounts for flight feasibility and safety constraints. The GWO algorithm is designed to efficiently explore the configuration space of UAVs, aiming to minimize the defined cost function and determine an optimal path. Simulations demonstrate the effectiveness and advantage of the proposed approach. Furthermore, the generated paths are validated through practical experiments, confirming their applicability for real UAV operations. This study underscores the potential of GWO for efficient and safe path planning in complex bridge inspection scenarios.

Keywords: Autonomous Vehicle Navigation, Sensor Fusion, Multi-Modal Perception
Abstract: Earth dam outlet works are giant marvels of engineering and critical to a nation’s infrastructure. To maintain assured operation, the structure should be inspected periodically to preempt failures. Deep underground, the hazards associated with inspecting these structures are gravely consequential. This spurred the creation of the DamBotTM research program to create a robotic platform capable of performing inspections robotically. Localizing a robot in an earth dam outlet is a challenge for both navigation and mapping. This paper will describe these challenges in detail along with the software stack used by the DamBotTM to overcome them.

Keywords: Hardware Platform, Sensor Fusion, Vision Systems
Abstract: A long drive in the summer heat to an inspection site miles from civilization can be hard on any team of researchers, but doubly so for the equipment. It is not enough to demonstrate the proof of concept in the lab if the technology cannot deliver in the field. The team at the US Army Corps of Engineers has spent the last few years developing and deploying the DamBot to earth dam outlet Works and DoD test ranges all over the United States. Not every deployment was successful, but each time new lessons were learned. This paper will attempt to capture that wisdom so that it might act as a lessons-learned for other teams attempting to realize their own robotic systems operating in the real-world environment.

Keywords: Multi-Robot Systems, Motion and Path Planning, Machine Learning
Abstract: Landing an uncrewed aerial vehicle (UAV) on a moving surface vehicle in open waters is a formidable challenge due to the dynamic and complicated nature of the environment. The UAV is at risk of damage during the touch-down process, as the oscillatory motion of the surface vehicle can lead to hard impacts or tip-overs. In response, this paper introduces a trajectory optimization strategy for quadrotor for tracking moving surface vehicle using visual feedback from an onboard camera. The visual data is used for real time forecasting of the future motion of the surface vehicle. This prediction leverages a deep learning approach to determine the landing period in future when the surface vehicle's roll and pitch will fall below the critical landing threshold for a certain period, thereby providing a safe window for the impending landing. Our proposed method utilizes visual data from onboard camera on UAV only, obviating the need for external communication between UAV and surface vehicle. The proposed trajectory optimization method has been extensively tested in simulation environment and able to follow the target travel up to 5m/s even with highly maneuvering path. The proposed surface vehicle motion prediction was tested with recorded data and able to achieve the prediction error less than 3 degrees, and in real-time manner.

Keywords: Human Factors and Human-in-the-Loop, Human Interface, Systems for Service/Assistive Applications
Abstract: This paper proposes a discriminant method whether a person pays attention to something consciously (intentionally) or sees it unconsciously by using a time-series of gaze rate distributions. The gaze rate is defined as a statistical possibility of when and where many people look, e.g. in some target movie. In this paper, first, eye movement data from multiple participants wearing a head mount display (HMD) are measured when some movie is displayed. Second, the gaze rate is calculated as a distribution of statistical probabilities at each frame (time). Lastly, verification experiments of our proposed gaze rate-based conscious/unconscious discrimination have been conducted compared to results from saliency-map-based prior method, and have got better results numerically. The author think it is quite important and meaningful to distinguish whether a person look at something consciously or just sees unconsciously, because it is not difficult to understand such discrimination improves analysis of persons’ potential consciousness more precisely and more in detail.

Keywords: Haptics and tactile sensors, Systems for Service/Assistive Applications, Virtual Reality and Interfaces
Abstract: This paper models and simulates a solenoid-based force presentation device and presents the feasibility of pulling presentation. A weight is attached to the end of the solenoid plunger, and the other is connected to a spring. When the solenoid is activated, the weight is attracted to the solenoid, and when the current is cut off, the weight is returned to its original position by the spring. By properly designing the current in the coil, the motion of this weight can be made asymmetric acceleration motion, and pulling force can be presented. In order to confirm this feasibility, first, the proposed system is modeled. Next, the current was adjusted so that the solenoid’s pulling force could be expressed as a function of the current in the coil. The result of the simulation is that the generation of asymmetric acceleration motion of the weights was confirmed. Therefore, the feasibility of the proposed method was confirmed.

Keywords: Systems for Field Applications, Machine Learning, Systems for Service/Assistive Applications
Abstract: In the context of the livestock industry, sheep hold significant economic value. The economic losses incurred from the loss of a few sheep pose an unacceptable burden for small and medium-sized ranches. And manpower and time must be invested in locating lost sheep. This research focuses on the development of an application designed to aid in finding lost sheep. To save manpower and time, the use of drones is considered. The economy of small drone and their mobility offer an effective solution to this challenge. Machine learning techniques are employed for sheep recognition. Also, the research discusses the system, environmental conditions, and several approaches that contribute to a system’s execution speed and accurate execution to the recognition target.

Keywords: Mechanism Design, Mechatronics Systems, Human Factors and Human-in-the-Loop
Abstract: Walkways are basically flat because they are built to be easy for pedestrians to walk on, but there are small uneven surfaces such as Braille blocks for the visually impaired, drainage gutter grating, etc. Ordinary pedestrians are not affected by these small irregularities when walking, but in the case of wheeled mobile robots, these small irregularities have a negative impact on their running. In particular, small robots that are designed to operate in spaces where humans coexist tend to have small wheel diameters and often use casters for the follow wheels. Therefore, mobile robots are required to detect those small uneven surfaces and design the robot's path appropriately. However, these small uneven surfaces are difficult to detect with the accuracy of laser rangefinders commonly used in outdoor mobile robots. Therefore, this paper proposes a method to estimate these small uneven surfaces using the intensity information of reflected light from a laser rangefinder (LRF). Furthermore, we propose a method for designing a robot path that is less affected by the estimated uneven surfaces using the proposed method, and confirm the effectiveness of the method through experiments on our prototype robot.

Keywords: Medical Systems, Mechanism Design
Abstract: In this study, to address the limitations of box trainers and VR training systems, a wire-driven surgical training device was designed to assist surgeons in acquiring laparoscopic surgical skills. The tool elevates the remote center of motion (RCM) of the robot to emulate its position on the skin surface and ensure sufficient space beneath the stomach, thus providing a large space for surgery. In addition, the proposed device provides realistic motion guidance for novice laparoscopic surgeons by allowing the insertion of commercially available surgical tools that they are familiar with, thus making surgical practice convenient. Moreover, this system could be modified to act as a platform for real-time robotic laparoscopic surgery.

Keywords: Rehabilitation Systems, Control Theory and Technology, Mechanism Design
Abstract: In recent years, manned bases have begun to make extended stays in space and on other planets. Medical problems such as bone and muscle loss have been observed during long stays in space, and training equipment has been developed for use in space to improve these problems. In this study, we proposed a dynamic training method to maintain anti-gravity muscles under microgravity environment, aiming at the application to the long-duration stay in space including exploration of other planets. The training method focused on the function of antigravity muscles. This device is designed to activate the subject's antigravity muscles and achieve antigravity muscle maintenance in a microgravity environment by causing disturbance and instability in the subject's posture and making the subject conscious of maintaining an upright posture. In a previous study, we proposed a static training method to maintain posture in an anterior and posterior leaning posture. Therefore, this paper proposes a dynamic antigravity muscle training method that tilts forward and backward, and posture maintenance experiments were conducted using a prototype machine to realize this method.　The target movements are dynamic holding of forward and backward leaning postures. The EMG potentials of each muscle were measured and compared during the experimental movements in the static posture holding evaluation test conducted in the previous study. The results showed that the proposed method was superior as a training effect because the muscle activity of the gluteus maximus and rectus femoris muscles increased approximately 10-fold compared to the muscle potential results of previous studies.

Keywords: Building Automation, Automation Systems, Mechatronics Systems
Abstract: Increasing global population and urbanization pose challenges in the construction industry due to limited resources. Adaptive buildings introduce the field of ultralightweight structures and promise a 50% resource reduction, while doubling the building’s lifespan. They comprise sensors, actuators and a control unit and can actively counteract external disturbances. This paper presents an automation concept for such adaptive high-rise structures. By collecting the measurements of redundant sensors (e.g. strain gauges, position and pressure sensors) as well as of an optical measurement system, reliable estimation of the structural state is ensured, even in case of faults. The sensor information is processed in a cascaded control loop including decentralized programmable logic controllers (PLC). The computed control forces are applied to the structure by multiple hydraulically driven actuators integrated into the structural elements. An Ethernet-based network communication system ensures the reliable transfer of high amounts of data between the PLC units and a wear monitoring system, that continuously monitors the structural wear. The proposed automation concept is implemented exemplary for the first adaptive high-rise building world-wide and shows the potential to actively compensate static loads and damp structural vibrations, while ensuring the continuous and safe operation of the system.

Keywords: Network Systems, Hardware Platform, Plant Engineering
Abstract: This paper presents an investigation into the improvement of security and operation time in homomorphically encrypted systems using Field Programmable Gate Array (FPGA) technology. The primary objective is to generate keys efficiently, minimizing key sizes while maintaining security. By leveraging FPGA capabilities for key generation and key switching, smaller ciphertext sizes can be achieved, ultimately improving operation time. The paper focuses on the development of a sensor data encryption system implemented on an FPGA board. The proposed approach enables simultaneous key generation and encryption of incoming sensor data using generated keys. The developed system implemented fixed-size random number generation and prime number checking in hardware, subsequently expanding these capabilities to produce arbitrarily sized prime numbers.

Keywords: Sensor Fusion, Systems for Field Applications, Control Theory and Technology
Abstract: In this research, we developed a highly accurate pedestrian tracking system using multiple Light Detection And Rangings (LiDARs). Tracking using multiple LiDARs effectively covers a broader range and prevents occlusion. However, since multiple LiDARs give multiple observations for one pedestrian，they must be integrated, especially in a crowded environment, where it is necessary to distinguish these from observations of multiple nearby pedestrians. To address this issue, a joint probabilistic data association filter (JPDAF) that can handle multiple pedestrians is effective. Although the center-of gravity positions observed by multiple LiDARs do not match perfectly due to calibration errors as well as observation of different surfaces of an object, in this paper, it is shown that JPDAF accommodates multiple center-of-gravity positions and considers the overlapping of tracked objects, thus achieving more accurate tracking of approaching pedestrians. Experimental results show that real-time tracking of multiple pedestrians successfully integrates multiple observations generated from each pedestrian.

Keywords: Sensor Networks
Abstract: With the increasing demand for human motion monitoring in daily life, it puts forward higher requirements for the efficiency and accuracy of the calibration method of inertial measurement units. We previously proposed an inclination-based magnetometer linear calibration (I-Calibration) method, which eliminates the need for high-precision turntable and calculation process does not need to converge. The effectiveness of the I-Calibration method applied in the designed 12-observation scheme has been verified, but the effectiveness in indoor multi-interference environment during human movement are still needed to verify. Therefore, this paper designed a series gait movement experiment in an indoor environment as the sensors are placed on the knee and ankle joint of the participant. The results demonstrated that the I-Calibration can effectively calibrate the magnetometers during straight round trip and square path, as the findings in this study expands the application range and application scenarios of the I-Calibration method and the applications of the inertial measurement units in daily life for human motion monitoring.

Keywords: Sensor Networks
Abstract: Monitoring blood pressure on a daily basis is crucial in preventing hypertension, which is a leading factor in the development of cardiovascular diseases. The conventional blood pressure measurement method involves the use of cuff-type sphygmomanometers, which can be uncomfortable; therefore, a comfortable, non-contact blood pressure measurement method is warranted. Therefore, we propose a non-contact blood pressure estimation method based on channel state information (CSI) obtained from general-purpose Wi-Fi devices. CSI is capable of multipath detailed identification, and many CSI-based non-contact biometric estimation methods have been proposed in recent years. In the proposed method, respiration and heartbeat signals are obtained from CSI, and blood pressure is estimated using random forest regression. Experimental results indicate that the proposed method can estimate blood pressure with an average absolute error of less than 5 mmHg compared to sphygmomanometer measurements. Thus, non-contact blood pressure estimation using a general-purpose Wi-Fi device is demonstrated to be feasible.

Keywords: Systems for Search and Rescue Applications, Machine Learning, Sensor Fusion
Abstract: Unmanned Aerial Vehicles (UAVs) hold immense potential for critical applications, such as search and rescue operations, where accurate perception of indoor environments is paramount. However, the concurrent amalgamation of localization, 3D reconstruction, and semantic segmentation presents a notable hurdle, especially in the context of UAVs equipped with constrained power and computational resources. This paper presents a novel approach to address challenges in semantic information extraction and utilization within UAV operations. Our system integrates state-of-the-art visual SLAM to estimate a comprehensive 6-DoF pose and advanced object segmentation methods at the back end. To improve the computational and storage efficiency of the framework, we adopt a streamlined voxel-based 3D map representation - OctoMap to build a working system. Furthermore, the fusion algorithm is incorporated to obtain the semantic information of each frame from the front-end SLAM task, and the corresponding point. By leveraging semantic information, our framework enhances the UAV's ability to perceive and navigate through indoor spaces, addressing challenges in pose estimation accuracy and uncertainty reduction. Through Gazebo simulations, we validate the efficacy of our proposed system and successfully embed our approach into a Jetson Xavier AGX unit for real-world applications.

Keywords: System Simulation, Automation Systems, Building Automation
Abstract: Heat waves impose health risks for the population. Evaporative cooling facades (ECF) can reduce the additional risks resulting from excessive heating of cities known as the heat island effect utilizing evaporation cooling. As a basis for effective use and development of short-term operation strategies for these facades, this paper elaborates a system of partial differential equations to model the dynamics of ECF. The system equations are based on the transport equation and derived utilizing energy balances. Qualitative feasibility of the phenomenological model is demonstrated.

Keywords: Multi-Modal Perception, Human Interface, Entertainment Systems
Abstract: Most extant studies that present haptic stimuli to the human body simultaneously with audiovisual contents for enhancing the emotional experience use vibration stimuli to the upper body. Vibratory stimuli largely induce arousal and do not lead to relaxing feelings. In our previous study, we investigated the emotional effects of presenting stroking stimuli to the external auditory meatus during listening to natural sounds that evoke relaxation. The results of the previous study suggested that the stroking stimuli to the external auditory meatus increased the joyfulness and pleasantness of the sounds. In this study, we investigated whether the emotional effects of the stroking stimuli could also be applied to classical music that induces relaxation. A user study involving twelve participants showed that, as in the previous study, the stroking stimuli enhanced pleasant feelings for non-musical sounds, whereas they enhanced unpleasant feelings for music. These findings help develop emotional haptic interfaces.

Keywords: Multi-Modal Perception, Sensor Fusion, Human-Robot/System Interaction
Abstract: Person attribute estimation is a task of great importance for a variety of real-world robotic applications. While the computer vision community has made impressive progress over the last decade, they often rely on the sole use of RGB images from surveillance datasets and large deep-learning models without considering real-time requirements. By contrast, mobile robotic platforms have to deal with restricted resources but are often equipped with RGB-D cameras, offering complementary modalities. This paper presents an approach to robustly estimate soft-biometric attributes from full-body RGB-D appearances of persons in the surroundings of a mobile robot. The effects of depth, RGB and RGB-D data as input are analyzed, taking into account runtime and resource requirements. On the robotic attribute dataset SRL, it is shown that the presented approach outperforms other state-of-the-art approaches by a large margin. Furthermore, real-time requirements are met when applied on an NVIDIA Jetson AGX Xavier or even on a mobile CPU only. Finally, a previous system for person detection, upper body orientation estimation, and posture classification is integrated to enable an even more comprehensive perception of persons in the surroundings of a mobile robot in one joint approach. The source code for training and application will be made publicly available on GitHub

Keywords: Multi-Modal Perception, Robotic hands and grasping, Haptics and tactile sensors
Abstract: Object pose estimation is important for robot manipulation. Tactile methods of pose estimation are time consuming, but can be more effective than visual methods. Pose estimation by tactile sensation requires an observation model and the collection of data samples in advance. We propose an efficient sample collection method to shorten the pre-experimentation time for generating observational models. The effectiveness of the proposed method was verified by simulation using a simple sensor model. The proposed method is more accurate than the conventional methods such as grid search and random sampling with the same number of samples.

Keywords: Multi-Modal Perception, Machine Learning, Automation Systems
Abstract: In deep learning, training with multiple modalities generally improves the learning performance. In robot motion learning, human knowledge information can be regarded as additional modalities and aids robots to perform their tasks. Previous studies have shown that annotation of action segments improves the generalizability of motion learning. However, manual segmentation causes less reliability because of its subjectivity and is time-consuming for large-scale datasets. Automatic segmentation by generative models, such as hidden Markov and predictive learning models, is an alternative approach to find the action segments and may be replaced with human knowledge information. In this study, we expand a predictive learning model using recurrent neural networks to automatic segmentation of robot behavior. We evaluate the prediction errors and update the estimated segmenting points in the direction of larger errors. We also assume that the robot behavior can be described as sequences of continuous time series with time-varying dynamics. To represent these dynamics, we allocated the learning model with parametric biases. We conducted a numerical experiment that simulates a time series of sensor information. Comparing with the hidden Markov model, the results showed that our predictive learning model estimated the segmenting points and reflected the dynamic properties of estimated segments.

Keywords: Vision Systems, Automation Systems
Abstract: Automating in-store logistics processes in the retail industry poses significant challenges for robot manipulators. Contrary to warehouses, retails stores are subject to customer actions, which can imply non-standard tidying of products. This paper addresses the problem of detecting, discriminating, and accurately estimating the 6 degrees-of-freedom (6-DOF) pose of individual products, even in unexpected positions such as fallen or wrongly placed objects. The trained object detection model successfully discriminated similar-shaped objects of different brands/types commonly found in convenience stores. The detection is used to initialized the object position while several possible orientations are explored by a Fibonacci Multi-Start method. The estimated pose is then refined by a multi-scale projective Iterative Closest Point (ICP). The evaluation of the complete 6-DOF pose estimation module revealed its consistent ability to converge to the correct pose, avoiding local optima and achieving sub-millimetric precision. A working demonstration is presented, showcasing a robot rearranging a convenience store shelf. The overall system demonstrated the ability to detect fallen objects, estimate their poses, determine suitable grasping directions, and execute successful grasps. Importantly, the system's feasibility with minimal human intervention was demonstrated, allowing easy addition of new objects by convenience store employees or other stakeholders.

Keywords: Vision Systems, Human-Robot/System Interaction, Sensor Fusion
Abstract: Understanding how human utilize space is important to design optimal space layouts and indoor robot navigation considering surrounding conditions. Tasks such as abnormal activity detection and human-robot interaction could also leverage the space utilization data to provide deeper information to human operators and control the robot in an intelligent space respectively. However, recognizing actions is time-consuming labor and space utilization is difficult to define. In this paper, we proposed a novel method to infer space utilization from human state recognition data as an open-set solution using the word vector model which could greatly reduce labor cost for space monitoring and provide better generalization. Through experiments with multiple participants, the effect of our method is validated.

Keywords: Systems for Field Applications, Machine Learning
Abstract: Tasks such as environmental surveys and disaster response using skid-steering or crawler-type mobile robots are in demand. Highly accurate power consumption estimation is essential to performing these tasks efficiently. Previous studies have focused on estimating power consumption for skid-steering and crawler-type mobile robots. However, no method has been developed that can be applied to complex topography without limiting the geology, considering skidding during turning, which is unique to skid-steer and crawler-type mobile robots. In this study, we propose a method for estimating the current consumption of a skid-steering-type mobile robot’s drive system that considers its skidding. Experiments on a real robot show the applicability of the proposed method, which estimates drive train current consumption by measuring the skidding as vibration data.

Keywords: Autonomous Vehicle Navigation, Systems for Field Applications, Machine Learning
Abstract: Global localization allows a robot to determine its own pose (position and orientation) within a known map of its environment without having any prior information about its initial pose, and it is essential for robust localization and recovery after localization failure. Previous works have realized global position estimation indoors using WiFi signal strength, even in visually and geometrically similar environments - with the robot's orientation being estimated by its change in position over time. By using directional antennas and modeling the angle-of-arrival of WiFi signals, this work proposes a method to estimate global orientation directly. This method can be used for single-shot global pose estimation or for initializing a Bayesian Estimator. sssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssss

Keywords: Machine Learning, Multi-Robot Systems, Intelligent and Flexible Manufacturing
Abstract: Realization of tasks involving interaction with humans and other robots is important for solving real-world problems. Cooperative transportation is one of the representative tasks. Control methods utilizing model-based reinforcement learning are promising for these tasks under consideration of uncertainty. One of the concerns is, however, the risk of stochastic behavior on the environment during the exploration for learning. Although conventional studies have suppressed undesired behaviors explicitly, the discussions to acquire the suitable model for safe planning are limited. This paper focuses on the fact that model-based reinforcement learning requires long-term prediction, and proposes a method for model learning that takes long-term prediction accuracy into account. Specifically, a new algorithm is developed based on a meta-optimized model learning based on long-term prediction accuracy. The effectiveness of the proposed method was verified through experiments on a cooperative transportation task with two manipulators with force sensors. Experimental results show that the proposed method reduces the worst-case scenario more than the conventional model that merely minimizes the prediction loss of the next step.

Keywords: Machine Learning, Autonomous Vehicle Navigation
Abstract: Extensive research has been conducted on autonomous navigation systems for last-mile outdoor transportation. A method in previous work utilizes semantic segmentation to classify driving environments into classes such as recommended traversable areas for autonomous navigation following traffic rules in walking space. However, the presence of shadows on the road surface negatively impacts the accuracy of semantic segmentation, consequently leading to failures in autonomous navigation. This paper proposes an approach to demonstrate shadow-robust semantic segmentation, employing a multi-head semantic segmentation model. By utilizing our proposed method, autonomous navigation following traffic rules in walking space was successful under conditions with shadows.

Keywords: Control Theory and Technology, Vision Systems, Machine Learning
Abstract: Why do Recurrent State Space Models such as PlaNet fail at cloth manipulation tasks? Recent work has attributed this to the blurry prediction of the observation, which makes it difficult to plan directly in the latent space. This paper explores the reasons behind this by applying PlaNet in the pick-and-place fabric-flattening domain. We find that the sharp discontinuity of the transition function on the contour of the fabric makes it difficult to learn an accurate latent dynamic model, causing the MPC planner to produce pick actions slightly outside of the article. By limiting picking space on the cloth mask and training on specially engineered trajectories, our mesh-free PlaNet-ClothPick surpasses visual planning and policy learning methods on principal metrics in simulation, achieving similar performance as state-of-the-art mesh-based planning approaches. Notably, our model demonstrates a faster action inference and requires fewer transitional model parameters compared to the state-of-the-art robotic systems in this domain.

Keywords: System Simulation, Mechatronics Systems, Control Theory and Technology
Abstract: Magnetic levitation is a promising technology for multiple types of industrial applications. The use of superconductors has various benefits over electrical magnets, like large levitation gaps, superior energy efficiency and an inherently stable system. The integration of a superconductorbased levitation module to an industrial handling system would allow a decoupling of motion system and load and a frictionless motion of the load. The drawback of these systems are the motion induced oscillations of the load due to the nonstiff coupling of superconductor and load. To develop control algorithms that counter-act oscillations a model is necessary to predict oscillations induced by motion as well as the effect of stabilizing actors. As the underlying physics of superconductors and magnetic forces are a research field of their own and computation algorithms that solve these equations are complex the modeling approach presented uses simple empirical models, based on the resonance frequencies of the system. Furthermore, experimental parameter identification is done and the resulting system equations are successfully validated.

Keywords: Intelligent Transportation Systems, Decision Making Systems
Abstract: When it comes to long-distance trips with battery electric vehicles, recharging the battery during the journey is a crucial factor which significantly impacts the overall trip duration. The Fixed Route Electric Vehicle Charging Problem consists of finding the time optimal charging strategy on a given route. This is a complex challenge necessitating the consideration of battery constraints, heterogeneous charging infrastructure, and nonlinear charging characteristics. This study introduces an algorithm based on discrete dynamic programming taking into account all those factors. To assess its efficacy, a comparative analysis is conducted against a solver based on mixed-integer linear programming that assumes linear charging behaviors. The findings show a superior performance of the proposed discrete dynamic programming approach in terms of generating better charging strategies.

Keywords: Multi-Robot Systems
Abstract: The development of autonomous soccer robots has engaged numerous researchers around the world and poses a wide variety of challenges in not only encompassing mechanical components but also control strategy, where predicting the adversarial action plays a pivotal role. In this paper, we present our interception prediction approach based on the tree data structure called “interception tree”. Each pair of connected nodes represents an adversarial relation, where we predict the child optimal interception on the parent’s trajectory. The tree gives out the kick plan’s risk factor and is used for further strategy guidance in the shoot, pass and defense tactics. We demonstrate the effectiveness of our algorithm in terms of kicking and team performance through our 1200 simulated matches result over 6 set-ups which combine the different formations and defense strategies.

Keywords: Multi-Robot Systems, Systems for Field Applications, Automation Systems
Abstract: The advancement of swarm robotics has made tremendous strides in recent years, expanding the scope of its deployment from outdoor settings like construction sites and planetary exploration to indoor scenarios such as transportation-related activities. However, due to the large deployment of units in swarm robots, they cannot be engineered to have the costly and high-performance characteristics of conventional large robots. Thus, it is desirable to maintain low performance per unit and to reduce manufacturing costs. This paper presents a stochastic distributed exploration algorithm that accounts for the above features of swarm robots, and its performance is verified through both simulations and experiments. The proposed algorithm is based on a random walk and avoids path planning and high-precision sensing, enabling it to function well even with low-performance robots, relying only on the distance to the center of the search area based on random walks. Simulation outcomes demonstrate that the proposed algorithm can explore the search area with a specified exploration distribution. The experiments show that the robot can sequentially explore multiple search areas.

Keywords: Human Interface, Human-Robot Cooperation/Collaboration, Motion and Path Planning
Abstract: In this paper, we propose a new approach using remote operation for pick-and-place tasks in logistics environments. We have applied our previously proposed AI, which simultaneously optimizes skill execution order and robot movements, to a remote operation robot for logistics purposes and constructed a system that allows the robot to be operated with only simple remote instructions. This has allowed us to construct a system where, with only simple instructions like pick and place positions and desired skills, the robot can perform complex tasks by planning both the execution order and movements. Experiments conducted in a warehouse environment using this system showed that the task time could be reduced, and intuitive instructions could be given compared with the method of directly operating the robot.

Keywords: Multi-Robot Systems, Automation Systems, Control Theory and Technology
Abstract: This paper presents the design and implementation of a self-reconfigurable V-shape formation controller for multiple unmanned aerial vehicles (UAVs) navigating through narrow spaces in a dense obstacle environment. The selection of the V-shape formation is motivated by its maneuverability and visibility advantages. The main objective is to develop an effective formation control strategy that allows UAVs to autonomously adjust their positions to form the desired formation while navigating through obstacles. To achieve this, we propose a distributed behavior-based control algorithm that combines the behaviors designed for individual UAVs so that they together navigate the UAVs to their desired positions. The reconfiguration process is automatic, utilizing individual UAV sensing within the formation, allowing for dynamic adaptations such as opening/closing wings or merging into a straight line. Simulation results show that the self-reconfigurable V-shape formation offers adaptability and effectiveness for UAV formations in complex operational scenarios.

Keywords: Human Factors and Human-in-the-Loop, Modeling and Simulating Humans, Supply Chain Management Systems
Abstract: The ongoing transition from Industry 4.0 to Industry 5.0 brings promising prospects for small and medium-sized enterprises (SMEs) to bolster their competitiveness and sustainability. To fully capitalize the benefits brought by this transition, SMEs need to master the upskilling of their workforce that conforms to the rhythm of technological development. Creating an educational symbiosis which supports this upskilling prompt is therefore crucial. The interdependence between research and education institutions and SMEs plays a vital role in cultivating a culture of continuous learning. This enables SMEs to effectively tackle the hurdles posed by the ongoing shift from technology-driven competition to one centered around creativity. The authors explore the European projects undertaken in recent years, emphasizing the insights that have played a pivotal role in empowering SMEs to fully leverage the potential of technological advancements within Industry 5.0. Furthermore, observations concerning the collaboration between SMEs and research and educational institutions have also been incorporated.

Keywords: Logistics Systems, Supply Chain Management Systems, Intelligent and Flexible Manufacturing
Abstract: This work is a case study approach to optimizing warehouse operations for the oil and gas industry. As companies in the energy industry seek to decrease costs, optimize industrial processes, and achieve higher degrees of sustainability to cut emissions, new automation tools are considered for utilization. The oil and gas industry is under scrutiny from governments, citizens, and activist organizations due to high carbon emissions, and need new initiatives across business areas that can contribute to decreases in emissions without compromising energy supply. This paper presents four different scenario setups of a spare parts warehouse for an offshore oil production installation in the Barents Sea outside the Norwegian coast. The scenarios are created in simulation software according to real-life dimensions. Various automation tools are used in the different scenarios to observe the degree to which time expenditure is affected in each scenario. They are compared according to the average time spent on order picking, consolidation, and shipping. Conclusively, we observe that a combination of few automation tools and a high degree of interoperability make for the most efficient oil and gas spare parts warehouse. For further work, we suggest cost and sustainability analyses.

Keywords: Rehabilitation Systems, Integration Platform, Welfare systems
Abstract: This paper presents a versatile automated gait lab system architecture to solve the challenges of collaborating and communicating among multiple instruments operated in gait analysis and rehabilitation. Our results demonstrate that the system architecture efficiently facilitates collaboration and synchronization among multiple instruments with varying update frequencies, enabling the successful implementation of closed-loop robotic exoskeleton assistance.

Keywords: Human Factors and Human-in-the-Loop, Human-Robot/System Interaction, Virtual Reality and Interfaces
Abstract: This paper introduces an information presentation strategy for pedestrians, aiming to enhance traffic efficiency in a mixed pedestrian-automated vehicle environment, such as a public road. While automated driving technology has made remarkable progress, interactions with pedestrians on regular roads have mostly been studied in virtual environments using virtual reality goggles. According to these studies, potential traffic efficiency and safety issues arise from pedestrians’ limited understanding of automated vehicle behavior. To address this, we propose a human-machine interface employing a head-mounted display (HMD) to mitigate traffic efficiency degradation caused by pedestrians. The proposed system draws upon behavioral economics principles to encourage pedestrians to modify their behavior and develop better interactions with automated vehicles. Simulations were conducted to identify an information presentation strategy that strongly supports learning, and its effectiveness was further validated through experiments involving a real vehicle. Notably, the experimental results confirmed that the information presentation strategies proven effective in simulations also facilitated pedestrian learning during real-world interactions.

Keywords: Systems for Service/Assistive Applications, Welfare systems, Human-Robot/System Interaction
Abstract: The next years will put more and more focus on the application of service robots in elderly care, especially nursing homes, to meet the "elderly wave" and at the same time be able to provide high-level, individual care of the residents. However, new challenges appear when introducing service robots in an environment where residents with impaired cognitive skills, like dementia, coexist with the service robots. This, especially, due to the fact that the residents mostly cannot, and formally, are not responsible for themselves. Thus, any accident will per definition, be the robot's responsibility. Unfortunately, there is a lack of standards and best practice about how to establish a safe and harmonic environment where dementia people and service robots coexist and collaborate. Unlike industrial robot installations, both the physical and interperceptual risks must be assessed when introducing service robots to nursing homes. This, because fear can cause unwanted reactions from residents, which in turn can create new risks. This paper addresses the complexity of performing risk assessment for service robots operating in coexistence with dementia residents in nursing homes. Both physical and interperceptual risks are included. Based upon the traditional ISO analysis, an extended method is suggested to minimize the unavoidable, residual safety risk.

Keywords: Human-Robot/System Interaction, Modeling and Simulating Humans, Human Factors and Human-in-the-Loop
Abstract: This paper presents a psychologically inspired engineering approach about optimal robot behaviour in Human-Robot Interaction (HRI). The main contribution of this paper is to introduce the concept of Organic Human-Robot Interactions (O-HRI). First we build up a basic formula for behaviour that can serve as a ground for robot behaviour. Then we introduce some crucial psychological aspects to robot behaviour, namely mistakes and miscalculations, imperfections and intentions in behaviour. The paper argues that mere advancement of certain social skills in social robots are not enough to make them more sociable or socially desirable. Therefore psychological, ethological and evolutionary backgrounds of social behaviours are discussed as fundamental baselines for our proposed concept of Organic Human-Robot Interaction. Finally we suggest some practical guidelines to consider when building up the behavioural paradigms of social robots.

Keywords: Human Factors and Human-in-the-Loop, Intelligent Transportation Systems, Autonomous Vehicle Navigation
Abstract: Aiming for a society where humans and automated vehicles can coexist cooperatively, understanding what constitutes cooperative and trustworthy behaviour is essential to designing automated vehicle controllers that enable the integration of highly automated vehicles into the real world. This study investigates how merging vehicles can gain trust from human-driven vehicles in a congested merging situation that requires explicit and implicit communication. Specifically, this study examines how the different behaviours of merging vehicles in the preparatory phase of the merge affect perceived trust from the perspective of the host vehicle in the mainstream lane. The findings suggest that transparent longitudinal positioning could improve the chance of successful merging, and cooperative deceleration during merging preparation could enhance the trust perceived by the host vehicle. Furthermore, the results reveal that, in time-sensitive situations where the merging vehicle approaches a lane closing point, prompt and decisive action of the merging vehicle encourages establishing trust with the host vehicle; any delay or hesitation can result in a lower level of trust. The results can provide valuable insights towards developing collaborative automated vehicles that improve safety and efficiency in real-world traffic situations that involve humans.

Keywords: Human Factors and Human-in-the-Loop, Human-Robot/System Interaction, Human-Robot Cooperation/Collaboration
Abstract: This paper discusses the relationship between robot movements and sense of agency in the teleoperation of a robot arm using a graphical user interface. Based on the sense of agency model, this study focuses on the imagery of robot movements held by operators. It verifies the relationship between imagined robot movements and preferred actual robot movements, as well as the relationship between robot movements and the sense of agency during operations, through two experiments. These findings suggest that operators exhibit a bias toward favoring swifter motions than those depicted in the imagery. Additionally, this study suggests that when there are significant differences in the movement distance and speed between movements based on imagined robot actions and those that consider bias, the sense of agency decreases. In particular, when considering the bias towards preferring faster movements, adjusting the control to achieve the same acceleration as movements that consider this bias, even when the movement distance changes, can evoke a sense of agency during the operation.

Keywords: Sensor Fusion, Motion and Path Planning, Mechatronics Systems
Abstract: Abstract— Autonomous navigation, steadily growing in prominence, offers transformative solutions ranging from simplifying repetitive tasks to ensuring safety in hazardous environments. In this research, we delve into advanced localization strategies, commonly denoted as 'slip compensation' techniques. By systematically analyzing primary localization approaches, we ascertain the most efficacious integrations and real-world implementations. Our investigation underscores inherent challenges, points of divergence, and significantly, emergent methodologies that enhance system resilience. Through our comprehensive findings, we illuminate pivotal facets of autonomous navigation, differentiating between indispensable elements and secondary factors, thus paving the way for future innovations in the field.