1
Concepts and Foundations Automation and Emerging Technologies

1.1 Introduction

The term "automation" generally refers to a set of automated processes on an input based on expert-inspired thinking. In a broader and more specialized field of management, the term includes automatic data retrieval (DR) and intelligent digital data processing (DP) to understand and quantify the situation. The input data to these systems are a set of numbers that are transmitted online or offline using advanced hardware and even a single image processor. Figure 1.1 shows a simple structure of an automated system for determining road conditions (at Amir Kabir University of Technology, Tehran Polytechnic), which is used for the automation in infrastructure management. Automation encompasses a range of activities from information extraction to robotic navigation, data analysis, solution presentation and knowledge discovery, and knowledge learning and self-learning.

Data on the condition of an infrastructure, such as roads, bridges, tunnels, railroad tracks, or a microscopic image of a bitumen mixture automatically with the machine, are first converted into digital format and stored as input in computer memory or transferred online to the analyzer software. These digital data can be processed or displayed and controlled simultaneously or on a high-resolution monitor. In general, the process of digitization of road infrastructure scans includes all operations of digitization, storage, processing, and display of output through the computer. The program inputs are then transferred to the processor after storage or online through the terminal. After processing the outputs through the same terminal, the data are available and usable. Figure 1.2 shows the chain of automation steps for automating typical processing.

Automated processing and automation have a wide range of applications, such as automatic assessment of road surfaces, automatic assessment of bridges and technical structures, assessment and inspection of tunnels, evaluation of pavement texture roughness, quality control of pavement markings, road safety audit, detection of signs, and classification. For this purpose, advanced devices in-line with modern technology such as remote sensing (RS), use of satellites and other spacecraft, robotics and automatic inspection, and advanced laser equipment are used. The data obtained by the multipurpose usability, such as images, are used for the simultaneous assessment of pavement distress as well as road audits.

This data may be used to isolate and monitor the health of infrastructure or to diagnose damage or other infrastructure characteristics, such as surface drainage, friction, roughness, and slope and arch determination. Images captured by automated systems are utilized as important data to detect various types of failures or to visually evaluate decisions. Figure 1.3 provides examples of several different types of images in infrastructure. Other needs and applications of automation range from robot insights for automation in aerial road imaging to the movement of ligaments on tunnel wall bodies that fall into the automation category. In other words, whenever a machine receives two- or higher-dimensional data, an image is eventually processed. Although there are many methods and limitations to image processing, in this text, we will consider the following basic classes:

Figure 1.1 An example of a schematic automatic robotic information retrieval system for automation in infrastructure management.

Figure 1.2 The schematic chain of automation steps for automating typical processing.

• Structure and framework of automation and key performance indices (KPIs)
• Advanced image processing techniques
• Fuzzy techniques and recent advances
• Automatic detection and its applications in infrastructure
• Feature extraction and fragmentation methods
• Feature prioritization and selection methods
• Classification methods and their applications in infrastructure management
• Models of performance measures and quantification in automation
• Nature-Inspired Optimization Algorithms (NIOAS)

Figure 1.3 Examples of 2D images as an input for the processing step. (a) Pavement without surface damage with coarse texture, (b) pavement with small transverse cracks, (c) pavement with bitumen damage in the path of wheels, (d) pavement with high surface drainage capability, (e) pavement with rutting distress in wheel path, and (f) pavement with alligator-type surface cracking distress (fatigue cracking).

1.2 Structure and Framework of Automation and Key Performance Indexes (KPIs)

Automation is a completely systematic process that requires basic design. If any of the steps are designed incorrectly or the process is not followed correctly, it may lead to the failure of the use of automation in the future. Given the importance of this issue, the elements of infrastructure management and its automation should be considered. The three main components in automation design are the following:

• Include data retrieval (DR)
• Data processing (DP)
• Data and information (DI) interpretation

Figure 1.4 shows the general architecture used to design the automation. A discussion of the general structure and macroarchitecture of the development of infrastructure models at the network level as well as the main components of the overall system design modules is given in Chapter 2.

Figure 1.4 General architecture used to design the automation.

The initial selection of indicators and the acceptance of these indicators play an important role in the implementation of automation and its success. As a general principle, the selection of this module directly affects the selection of DR because indicators are ordered according to need. The collection of certain information requires the use of special equipment.

In national and macroautomation systems, the choice of technology depends a lot on the indicators desired by managers and affects the level of management. For this reason, the order of the automation chain is different in practice and requires a top-down design. Before designing any system, it is necessary to have a proper understanding of the types of common indicators in the management of roads and technical buildings. After fully understanding the needs, then the role of emerging technologies and future research of automation largely affects the selection of the method.

1.3 Advanced Image Processing Techniques

In order to evaluate and analyze the images, it is often necessary to extract directional information on the subjects (including cracking, texture, aggregate morphology, morphology of bitumen contents, friction, etc.) in the image. For this reason, multilevel methods are considered as efficient tools due to their ability to decompose information in several levels and the possibility of reconstructing them with the least amount of error.

In this section, various types of single and multilevel methods are introduced, then, using the indicators introduced in Section 1.2, the efficiency of each multilevel method in specialized issues is evaluated. The main characteristics of image quality evaluation, as shown in Figure 1.4, are evaluated for each method and with different filters.

Figure 1.5 General steps of digital processing in automation.

The three main components for any automated system are the following: (i) image capture device, or the image equitization component (IAC); (ii) image analysis system and related algorithms, or the image processing component (IPC); and (iii) interpretation and indexing methods, or the image interpretation component (IIC). To perform analysis on the image, six general steps are required to obtain the result, including preprocessing, segmentation, feature extraction, feature selection, detection, and classification.

One of the most important methods of improving data quality is multilevel analysis that consists of the wavelet approach (WA), curvelet approach (CA), ridgelet approach (RA), and shearlet approach (SA), which has many applications. In this section, the capability of each of these single-level and multilevel methods in noise elimination is clearly presented, then the types of filters and optimal filter selection methods are introduced. Finally, the optimal method is selected from the existing methods as an example for case studies in the field of pavement management according to the capabilities of the method (see Figure 1.5).

The various preprocessing methods presented in Chapter 3 are used to improve the quality of images, with the aim of removing noise and enhancing the image in order to increase the detection power in the separation and detection stages via image processing. Also, different single-level and multilevel methods will be studied in detail. In general, multilevel methods work better than single-level methods. Among the multilevel methods, the complex Shearlet...