From the Publisher: 
The accessible presentation of this book gives both a general view of the entire computer vision enterprise and also offers sufficient detail to be able to build useful applications. Users learn techniques that have proven to be useful by first-hand experience and a wide range of mathematical methods. A CD-ROM with every copy of the text contains source code for programming practice, color images, and illustrative movies. Comprehensive and up-to-date, this book includes essential topics that either reflect practical significance or are of theoretical importance. Topics are discussed in substantial and increasing depth. Application surveys describe numerous important application areas such as image based rendering and digital libraries. Many important algorithms broken down and illustrated in pseudo code. Appropriate for use by engineers as a comprehensive reference to the computer vision enterprise.

Computer vision : a modern approach = 计算机视觉 : 一种现代的方法

비만아 부모의 돌봄 경험: q 방법론

Applications of machine learning to machine fault diagnosis: A review and roadmap

1D convolutional neural networks and applications: A survey

During the last decade, Convolutional Neural Networks (CNNs) have become the de facto standard for various Computer Vision and Machine Learning operations. CNNs are feed-forward Artificial Neural Networks (ANNs) with alternating convolutional and subsampling layers. Deep 2D CNNs with many hidden layers and millions of parameters have the ability to learn complex objects and patterns providing that they can be trained on a massive size visual database with ground-truth labels. With a proper training, this unique ability makes them the primary tool for various engineering applications for 2D signals such as images and video frames. Yet, this may not be a viable option in numerous applications over 1D signals especially when the training data is scarce or application-specific. To address this issue, 1D CNNs have recently been proposed and immediately achieved the state-of-the-art performance levels in several applications such as personalized biomedical data classification and early diagnosis, structural health monitoring, anomaly detection and identification in power electronics and motor-fault detection. Another major advantage is that a real-time and low-cost hardware implementation is feasible due to the simple and compact configuration of 1D CNNs that perform only 1D convolutions (scalar multiplications and additions). This paper presents a comprehensive review of the general architecture and principals of 1D CNNs along with their major engineering applications, especially focused on the recent progress in this field. Their state-of-the-art performance is highlighted concluding with their unique properties. The benchmark datasets and the principal 1D CNN software used in those applications are also publically shared in a dedicated website.

1D Convolutional Neural Networks and Applications: A Survey

This paper investigates a robust fault-tolerant control scheme for uncertain robot manipulators. The proposed scheme is designed via active fault-tolerant control method by combining a fault estimation scheme with a novel robust adaptive quasi-continuous second-order sliding mode (AQC2S) controller, so as to accommodate not only system failures but also uncertainties. First, a neural network based fault estimation is designed to online approximate the unknown uncertainties and faults. The estimated uncertainty and fault information are then used to compensate in advance for the effects of uncertainties in fault-free operation and both uncertainties and faults in fault operation. To eliminate the neural network compensation error, QC2S with adaptation gain, named as adaptive QC2S (AQC2S), is proposed. By integrating the advantages of the neural network observer and the AQC2S controller, the integrated scheme has a good capability to accommodate both the uncertainties and faults with chattering-free, higher position tracking accuracy, and no requirement of prior knowledge of the fault information. The stability and convergence of the proposed fault-tolerant control system is proved theoretically. Simulation results for a PUMA560 robot demonstrate the effectiveness of the proposed algorithm.

Robust fault-tolerant control for uncertain robot manipulators based on adaptive quasi-continuous high-order sliding mode and neural network

In this study, a novel Fault-Tolerant Control Methodology (FTCM) is developed for robot manipulators. First, to overcome singularity glitch and to enhance convergence time of conventional Terminal Sliding Mode Control (TSMC), a new Fast Terminal Sliding Mode Surface (FTSMS) is constructed. Next, to reduce the computation complexity and to provide requirements about undefined nonlinear functions for the control system, a Disturbance Observer (DO) to estimate uncertain dynamics, external disturbances, or faults. Besides, a Super-Twisting Reaching Control Law (STRCL) is designed to compensate for the estimated error of disturbance observer with chattering rejection. Final, a novel, robust, FTCM was developed for robot manipulators to obtain the stability goal of the system, to reach the prescribed performance, and to overcome the effects of disturbances, nonlinearities, or faults. Accordingly, the proposed FTCM has remarkable features, such as fast convergence speeds, robust precision, high tracking performance, significant alleviation of chattering behavior, and finite-time convergence. The position tracking computer simulations were implemented to exhibit the effectiveness and feasibility of the suggested FTCM compared with other control algorithms.

/pdf/a-novel-fault-tolerant-control-method-for-robot-manipulators-1yflfc2n0f.pdf

A Novel Fault-Tolerant Control Method for Robot Manipulators Based on Non-Singular Fast Terminal Sliding Mode Control and Disturbance Observer

This paper presents a novel method for fusing information from multiple sensor systems for bearing fault diagnosis. In the proposed method, a convolutional neural network is exploited to handle multiple signal sources simultaneously. The most important finding of this paper is that a deep neural network with wide structure can extract automatically and efficiently discriminant features from multiple sensor signals simultaneously. The feature fusion process is integrated into the deep neural network as a layer of that network. Compared to single sensor cases and other fusion techniques, the proposed method achieves superior performance in experiments with actual bearing data.

/pdf/a-deep-neural-network-based-feature-fusion-for-bearing-fault-63usgpbz7q.pdf

A Deep Neural Network-Based Feature Fusion for Bearing Fault Diagnosis.

A new control scheme is proposed for motion tracking of a Takagi–Sugeno fuzzy system using the backstepping quasi-continuous high-order sliding mode (HOSM) control technique. First, a Takagi–Sugeno fuzzy model is used to represent the original second-order nonlinear system; most of the parameters for this model can be computed offline. Next, a conventional backstepping sliding mode control (BSMC) is designed to stabilize and guarantee the exact motion tracking for the Takagi–Sugeno fuzzy system. However, use of the conventional sliding mode control generates significant chattering. Therefore, a quasi-continuous second-order sliding mode (QC2S) control is employed to reduce chattering and obtain higher tracking precision, resulting in a backstepping quasi-continuous second-order sliding mode (BQC2S) control law. Combining the Takagi–Sugeno fuzzy model with the BQC2S controller results in a controller scheme that preserves the advantages of both techniques, such as the low online computational burden of the Takagi–Sugeno fuzzy model, and the low chattering, robustness, and fast transient response of the BQC2S controller. Finally, the proposed controller is used to control a two-link robot manipulator and is compared with the existing approaches. Simulation results are presented to demonstrate the effectiveness of the proposed methodology.

Backstepping quasi-continuous high-order sliding mode control for a Takagi-Sugeno fuzzy system with an application for a two-link robot control

This paper develops a new method to control uncertain robot manipulators by using only position measurements. The controller is designed based on a combination of a computed torque controller (CTC) with a higher-order sliding-mode observer and a fuzzy compensator. First, three higher-order sliding-mode (SM) observers (second-order SM, third-order SM and third-order SM linear (TOSML) observers) are designed and compared to verify whether the TOSML observer is the best for observing velocity and identifying uncertainty. A combined CTC-TOSML controller was then designed. Although this controller scheme can overcome the drawbacks of conventional CTCs, its tracking performance can still be improved. To enhance capability of the tracking performance, a CTC-TOSML controller plus fuzzy compensator called a CTC-TOSML-Fuzzy controller is proposed. The proposed controller increases the potential of the CTC for real robot applications. Finally, computer simulation results on a PUMA560 robot are discussed to verify the effectiveness of the proposed strategy.

Hee-Jun Kang

Papers

Robust fault-tolerant control for uncertain robot manipulators based on adaptive quasi-continuous high-order sliding mode and neural network

A Novel Fault-Tolerant Control Method for Robot Manipulators Based on Non-Singular Fast Terminal Sliding Mode Control and Disturbance Observer

A Deep Neural Network-Based Feature Fusion for Bearing Fault Diagnosis.

Backstepping quasi-continuous high-order sliding mode control for a Takagi-Sugeno fuzzy system with an application for a two-link robot control

Output feedback tracking control of uncertain robot manipulators via higher-order sliding-mode observer and fuzzy compensator