In this paper, we propose a stage-wise knowledge transfer method that uses block-wise retraining to transfer the useful knowledge of a pre-trained residual network (ResNet) in a teacher-student framework (TSF). First, multiple hint information transfer and block-wise supervised retraining of the information was alternatively performed between teacher and student ResNet models. Next, Softened output information-based knowledge transfer was additionally considered in the TSF. The results experimentally showed that the proposed method using multiple hint-based bottom-up knowledge transfer coupled with incremental block-wise retraining provided the improved student ResNet with higher accuracy than existing KD and hint-based knowledge transfer methods considered in this study.

Optical character recognition (OCR) has been studied for decades because it is very useful in a variety of places. Nowadays, OCR's performance has improved significantly due to outstanding deep learning technology. Thus, there is an increasing demand for commercial-grade but affordable OCR systems. We have developed a low-cost, high-performance OCR system for the industry with the cheapest embedded developer kit that supports GPU acceleration. To achieve high accuracy for industrial use on limited computing resources, we chose a state-of-the-art text recognition algorithm that uses an end-to-end deep learning network as a baseline model. The model was then improved by replacing the feature extraction network with the best one suited to our conditions. Among the various candidate networks, EfficientNet-B3 has shown the best performance: excellent recognition accuracy with relatively low memory consumption. Besides, we have optimized the model written in TensorFlow's Python API using TensorFlow-TensorRT integration and TensorFlow's C++ API, respectively.

Convolutional Neural Networks (CNN) have been used extensively in recent times to solve image classification and segmentation problems. However, the use of CNNs in image super-resolution problems remains largely unexploited. Filter interpolation and prediction model methods are the most commonly used algorithms in super-resolution algorithm implementations. The major limitation in the above named methods is that images become totally blurred and a lot of the edge information are lost. In this paper, we analyze super resolution based on CNN and the wavelet transform super resolution method. We compare and analyze the performance according to the number of layers and the training data of the CNN.

This paper proposes vision sensors and deep learning-based around view monitoring system for ship berthing. Ship berthing to the port requires precise relative position and relative speed information between the mooring facility and the ship. For ships of Handysize or higher, the vesselships must be docked with the help of pilots and tugboats. In the case of ships handling dangerous cargo, tug boats push the ship and dock it in the port, using the distance and velocity information receiving from the berthing aid system (BAS). However, the existing BAS is very expensive and there is a limit on the size of the vessel that can be measured. Also, there is a limitation that it is difficult to measure distance and speed when there are obstacles near the port. This paper proposes a relative distance and speed estimation system that can be used as a ship berthing assist system. The proposed system is verified by comparing the performance with the existing laser-based distance and speed measurement system through the field tests at the actual port.

IIoT stands for Industrial Internet of Things used in manufacturing, healthcare, and transportation in networked smart factories. Recently, IIoT's environment requires an automated control system through intelligent cognition to improve efficiency. In particular, IIoT can be applied to automatic calibration of production equipment for improved management in industrial environments. Such automation systems require a wireless network for transmitting industrial data. Self-calibration systems in laser transmission paths using wireless networks can save resources and improve production quality by real-time monitoring and remote control of laser transmission path. In this paper, we propose a wireless networked system for self-calibration of laser equipment that requires a laser transmission path, and we show the results of the prototype evaluation. The self-calibration system of laser equipment measures the coordinates of the laser points with sensors and sends them to the host using the proposed application protocol. We propose a wireless network service for the wired motor controller to align the laser coordinates. Using this wireless network, the host controls the motor by sending a control command of the motor controller in an HTTP message based on the received coordinate values. Finally, we build a prototype system of the proposed design to verify the detection performance and analyze the network performance.

This paper introduces a method to control an industrial robot arm to imitate the movement of the human arm and hand using electromyography (EMG) signals. The proposed method is implemented on the UR3 robot that is a popular industrial robot and a MYO armband that measure the EMG signals generated by human muscles. The communications for the UR3 robot and the MYO armband are integrated in the robot operating system (ROS) that is a middle-ware to develop robot systems easily. The movement of the human arm and hand is detected by the MYO armband, which is utilized to recognize and to estimate the speed of the movement of the operator's arm and the motion of the operator's hand. The proposed system can be easily used when human's detailed movement is required in the environment where human can't work. An experiments have been conducted to verify the performance of the proposed method using the teleoperation of the UR3 robot.

In situations where mobile robots are operated either by autonomous systems or human operators, such as smart factories, priority-based teleoperation is crucial for the multiple operators with different priority to take over the right of the robot control without conflict. This paper proposes a priority-based teleoperation system for multiple operators to control the robots. This paper also introduces an efficient joystick-based robot control command generation algorithm for differential-drive mobile robots. The proposed system is implemented with ROS (Robot Operating System) and embedded control boards, and is applied to Pioneer 3AT mobile robot platform. The experimental results demonstrate the effectiveness of the proposed joystick control command algorithm and the priority-based control input selection.

This paper finds solutions for using ZigBee in wireless networked control system (WNCS). The round trip time delay and packet loss rate of the WNCS are measured. On the basis of these measured data, a playback buffer is used to solve the variable time delay in WNCS, and a Smith predictor is introduced to compensate for the time delay. The WNCS was able to be actually constructed to perform DC motor position control with 40 Hz sampling frequency.