In recent years, the oceans debris is increased because of human’s activity. Especially, the plastic debris, such as plastic bottles, is not biodegradable, and therefore those are creating serious environmental problems. The ocean debris collecting in the world is done through volunteer activities. However, these activities require a lot of time and labor. In this research, we proposed a method of ocean debris collecting USV operation method. Also, we designed and developed a USV (Unmanned Surface Vehicle) for the purpose of autonomous oceans debris collecting mission and evaluated the developed ocean debris collecting device. In the development of USV, electronic parts, which are to operate the mission autonomously, were selected. Then, each electronic part was placed inside the waterproof box according to the designed power and communication system diagram. In the development of oceans debris collecting device, we designed a belt conveyor type device. A motor was selected to rotate the device and a decelerator, which used the planetary gear mechanism, was designed. To evaluate the performance of the developed ocean debris collecting device, we tested various types of plastic bottles in tank environment. As a result of the experiment, the developed USV collected various types of plastic bottles with a 47.7% probability.

抄録全体を表示

A prototype honeypot system based on the Modbus/TCP protocol is designed for the protection of Industrial Control Systems (ICS). The proposed system operates under the control of a single server and enables multiple agents, each with several honeypot devices, to be deployed in different industrial environments. For each honeypot, the device characteristics are defined by JSON description files. The experimental results show that the interaction behavior of the proposed honeypot is closer to that of an authentic ICS device (a PLC) than that of the Conpot open-source ICS honeypot reported in the literature. Furthermore, the honeypot is awarded a perfect score by the honeypot scoring mechanism of Shodan Internet of Things (IoT) search engine

抄録全体を表示

A framework for discussing the neural underpinning of communication processes is proposed from the perspective of synchronization. This framework comprises four stages: (i) characterizing the target communication in a two-dimensional space defined by symbolic/embodied (non-symbolic) and voluntary/involuntary processes, (ii) focusing on the level of analysis of synchrony on an ontological hierarchy, (iii) constructing a neurocognitive model to simulate neural dynamics, and (iv) testing an empirical hypothesis on the neural underpinning of communication through model-based electroencephalography (EEG) connectivity neurofeedback in communication experiments with the cognitive neural mass/field model. We performed two EEG experiments, implementing the former two stages: the formation of symbolic communication, in which communication changed from voluntary to involuntary, and intentional switching in embodied communication, which involves switching between voluntary and involuntary behavior. The findings on communicative brain activities from these experiments culminated in the hypothesis that three brain regions are involved in interpreting symbols and motor intentions as well as in social coordination, in which one region might be shared by two modalities and the other two are specific to each modality. As we could perform the experiments and their analyses and derive a working hypothesis based on the framework, we claim that the proposed framework may be vital for investigating the neural underpinnings of communication in two different modalities in a unified manner.

抄録全体を表示

Due to the advent of Industrial 4.0, the Industrial Internet emphasizes the combination and application of IT technology and OT technology, and one of the features of Time sensitive networking is the feature of separating transmission time. This feature will be able to combine IT technology and OT technology in Industrial control systems. But as a result, the issue of information security in the industry is on the rise. In order to research security issue and protection in Time sensitive networking, we have built a test platform support TSN for industrial control networks for related research. Through the experiments in the research, we can know that TSN has a good effect on the separation of general traffic and TSN traffic. TSN won’t be affected by general traffic but is very vulnerable to priority traffic attacks.

抄録全体を表示

A driving roller arrangement of hemisphere is one of the important problems by omnidirectional sphere conveyance. In this research, the roller arrangement problem, viewed as an evaluation function, is thought of as mean of roller’s kinetic energy with respect to the sphere direction. Furthermore, theoretically, we calculate the evaluation function, and find the contact point such that the evaluated value is minimal.

抄録全体を表示

Industrial control system (ICS) and critical infrastructure have become increasingly dependent on cyber-physical systems nowadays. Since ICS network is vulnerable to adversarial attacks, building secure operation mode is essential. In order to secure the integrity of data in ICS, this paper proposes a blockchain-based log system implemented on physical industrial equipment. Applying cross-organizational blockchain transaction mechanism in the specialized master-slave network model, industrial control transmission logs can be verified and non-repudiation

抄録全体を表示

Conventionally, the brain signals were analysed manually by the neuroscientists on how the brain signals reacts in proportion with the human body. However, this process is very time consuming and unreliable. Therefore, we have proposed a brain signal detection system based on deep learning algorithm in response to the output of EEG device on the imagery finger movements. These fingers include thumb, index, middle, ring and little of right hand. In this study, 4 Convolutional Neural Network (CNN) classification models were developed. These 4 CNN models are different in terms of the pre-processing requirements and the neural network architecture. The best results for offline classification obtained in this project are 69.07% and 82.83% respectively in terms of average accuracy from 6-class and 2-class tests. Moreover, this project has also developed a proof of concept for applying the trained models in online or real-time classification.

抄録全体を表示

The use of the formal specification description language VDM++ in software design can eliminate ambiguity in the specification. However, software testing after implementation is necessary even if the design uses VDM++, but manually generating test cases is labor-intensive and time-consuming. Therefore, our laboratory developed BWDM, which is an automatic test case generation tool for VDM++ specifications. However, BWDM is not very useful because it has three problems about its narrow scope of application. This paper solves the three problems and improves the usefulness of BWDM by expanding the scope of application of VDM++ definitions and adding a function to generate test cases for object states. In addition, we conducted a comparison experiment with manual test case generation and confirmed that BWDM can reduce work time.

抄録全体を表示

In this paper, a fuzzy theory based identification system has been developed to improve the self-driving image recognition technology. The Raspberry Pi microprocessor is used as the main controller of the car. The author writes a Python program to deal with the image recognition problem. The image processing techniques include grayscale, binarization, morphology, image cutting and so on. There are four main functional tests in the scenario setting. It includes road identification, conversion of lane turning arc into front wheel turning angle, intersection identification, and traffic light identification. The purpose is to verify the functionality of fuzzy-based identified self-driving cars. The experimental results show that the developed recognition system based on fuzzy theory can successfully improve the recognition effect and reduce the probability of self-driving walking errors.

抄録全体を表示

Sign language is an important communication tool for deaf and hearing-impaired people. The study of sign language recognition can not only promote the communication between deaf-mutes and normal people, but also push the development of intelligent human-computer interaction. Sign language recognition based on deep learning has advantages in processing large scale dataset. Most of them use 3D convolution, which is not conducive to optimization. In this paper, an improved (2+1)D-ResNet model is proposed for isolated word recognition. The model convolves the video frame sequence in space and time dimensions and optimizes the parameters respectively. Based on CELU activation function, the accuracy of sign language recognition is improved effectively. The validity of proposed algorithm is verified on CSL dataset..

抄録全体を表示

Cross-view image matching for geo-localization is the task of finding images containing the same geographic target across different platforms. This task has drawn significant attention among researchers due to its vast applications in UAV’s self-localization and navigation. Given a query image from UAV-view, a matching model can find the same geo-referenced satellite image from the database, which can be used later to precisely locate the UAV’s current position. Many studies have achieved high accuracy on existing datasets, but they can be further improved by combining different feature processing methods. Inspired by previous studies, in this paper, we proposed a new framework by using a channel-based attention mechanism combined with a part-based representation learning method, including multi-level feature aggregation and an alternative pooling strategy to enhance the feature extracting process. The proposed model significantly improved matching accuracy and surpassed the existing state-of-the-art methods on University-1652 dataset..

抄録全体を表示

Machine vision has been widely used in the field of fruit picking in recent years, and its main application directions include fruit identification, fruit quality detection, fruit maturity detection and grading, etc. Among them, fruit maturity detection technology is of great significance for improving the quality and market competitiveness of fresh and stored fruits. This paper focuses on the application of machine vision in fruit identification, fruit ripeness detection and grading in the past three years, and the application is more mature in many fruits such as citrus, blueberry, cherry, etc. They use some algorithms to accurately identify the fruit, process its image, and feed it back to control the robotic arm for picking and other operations...

抄録全体を表示

In Japan, the population of working age between 15 and 64 years old peaked in 1995 at about 87 million and is expected to continue to decline in the future. Therefore, to solve the labor shortage, the introduction of industrial robots that can perform the same level of work as humans is strongly requested especially in the food preparation industry. In order to prevent danger to workers there, industrial robots must recognize workers and avoid them when there is fear of collision. In this paper, we propose a method for extracting hand regions based on the color distribution of a hand and GrabCut in an experimental environment to recognize human hands and detect their directions of approach. The proposed method was examined experimentally and gave satisfactory results.

抄録全体を表示

COVID-19 has now spread to every continent around the world. The virus is spread through the transmission of close contact with the infected person as well as by touching a surface that has been contaminated with the virus. Touchless switches must be utilized especially in office and home environments to reduce the spread of COVID-19. This paper presents the Smart Interactive Virtual Assistant for Smart Office Door Application. The proposed system uses voice recognition and artificial intelligence by applying the open-sourced API of Google Assistant to control the smart door contactless. The user can give the command to the system via voice to either open or close the door. This smart interactive voice assistant system uses Raspberry Pi as the brain or computer for the interface and Python is used as the software for running the coding script. The components used such as LED lights and DC motors are connected through the General-Purpose Input Output of Raspberry Pi and the relay module. One of the DC motors is used for controlling the lock and another one is used for controlling the door. Blue LED is used as a locked door notification while a green LED is used as an unlocked notification. A pair of microphones and speakers are also connected with the Raspberry Pi as input and output voice commands through a USB port and Bluetooth. With the smart interactive virtual assistant solution, office spaces will move towards a more touch-free control which could prevent the spread of COVID-19.

抄録全体を表示

With the popularization and advancement of digital technology and network technology in recent years, cyber security has emerged as a critical concern. In order to defend against malicious attacks, intrusion detection systems (IDSs) increasingly employ machine learning models as a protection strategy. However, the effectiveness of such models is dependent on the algorithms and datasets used to train them. The present study uses five different supervised algorithms (Naïve Bayes, CNN, LSTM, BAT, and SVM) to implement the IDS machine learning model. A data-balancing algorithm based on a generative adversarial network (GAN) is proposed to mitigate the data imbalance problem in the IDS dataset. The proposed method, designated as GAN-BAL, is applied to the CICIDS 2017 dataset and is shown to improve both the recall rate and the accuracy of the trained IDS models.

抄録全体を表示