**2. Industry 4.0: technologies for next-generation maritime logistics and shipping digitalization**

How does modern maritime logistics look like in the era of Industry 4.0? The concept of Industry 4.0, first formulated by the German government, mainly refers to the automation and digitization of manufacturing processes. Industry 4.0 encompasses cyber-physical systems2 , AI3 , IoT4 , cloud computing5 , cognitive computing6 , smart

<sup>2</sup> A Cyber Physical System is a system in which cyber space and physical space are more closely linked, in which information from reality (physical) is taken into a virtual space (cyber) by a computer, and the analysis results of the computer's computational power are fed back to derive optimal results in the real world. It is a system in which cyber space and physical space are more closely linked.

<sup>3</sup> AI, short name of Artificial Intelligence, is a branch of computer science that studies "intelligence" using the concept of "computation" and the tool of "computers."

<sup>4</sup> IoT, or Internet of Things, is a system in which various things (devices) are connected to the Internet to fully utilize the information those devices carried.

<sup>5</sup> Cloud computing is a form of usage in which computer resources are provided in the form of services via a computer network such as the Internet. It is sometimes referred to as cloud for short.

<sup>6</sup> Cognitive computing is a system in which computers not only process instructions given by humans, but also think and learn on their own like humans, and support decision-making.

#### *Shipping Digitalization and Automation for the Smart Port DOI: http://dx.doi.org/10.5772/intechopen.102015*

factories7 , and digital twin8 . Industry 4.0 has been implemented in many countries besides Germany, including the US, France, the UK, China, Japan, Korea, and Thailand. Particularly in the supply chain area, digitalization, integration of AI and IoT, sharing economy<sup>9</sup> , and BC are playing an important role in addition to these core technologies. Logistics has not been associated with high technology for a long time since it was recognized as an industry in the mid of nineteenth century. The situation is changing with the increase of Logistics 4.0 efforts, where innovative technologies such as AI, IoT, and BC are increasingly being implemented in the logistics industry, in parallel with similar efforts to many other industries, presenting even more efficient sustainability and human-centric approaches. However, despite the numerous regional initiatives, there are no adequate frameworks existing for companies in the logistics industry to embrace those technologies to the largest extent. There is a need for guidelines to implement new technologies in the logistics industry for the common good of the entire industry and eventually society as a whole.

Against this background, a closely related development, namely the Physical Internet (PI), has drawn attention from various parties as one of the most effective measures to improve logistics efficiency and reduce greenhouse gas emissions. The features of PI include interoperability, modularity, and standard interfaces and protocols. In order to take advantage of these features, technology to share data while maintaining confidentiality, such as Blockchain (BC) technology, is essential. This technology, also known as a distributed ledger, is difficult to tamper with, requires no administrator, and allows execution of smart contracts. The use of BC technology in PI is expected to dramatically accelerate the construction of a sustainable logistics network. In the following sections, we will present, in more detail, the potential of applying BC technology in the PI network context.

#### **2.1 AI**

Artificial Intelligence (AI) is a field of computer science that uses the concept of "computation" and the tool of "computers" to study and implement "intelligence."

AI can be classified into general-purpose AI and specialized AI. General-purpose AI is also referred to as strong AI. It is an AI that is not limited to a specific task but is capable of general-purpose processing; that is, it has the same intelligence as humans. Some examples of general-purpose AI are Doraemon and Astro Boy. None of them exist in the real world yet. Specialized AI is also called weak AI. It is an AI that specializes in performing a specific task. Image recognition, chess, Go, automated driving, human conversation, etc., are all examples of specialized AI.

In a broader sense, AI includes rule-based AI and machine-learning AI. Rulebased AI refers to making decisions according to rules described by humans. It also automates tasks that require hardware and human judgment. It can be described

<sup>7</sup> A smart factory is a highly productive and efficient factory that utilizes digital technologies such as AI and IoT.

<sup>8</sup> Digital twin is a technology that reproduces various data collected from the real world on a computer, as if they were twins. Based on the huge amount of data collected, the computer can perform physical simulations that are as close to reality as possible, which is an effective way to improve the manufacturing process of your products and services. Digital twin ship is a recognized prospect in shipping sector that helps optimize fleet management and enhance port and terminal operations.

<sup>9</sup> The sharing economy is a new economic movement in the form of renting, buying, selling, and offering among individuals via an Internet platform.

as an office robot. In contrast, machine-learning AI generally does not require a human to write the rules. It has algorithms for self-learning in machine-learning models, and it behaves intelligently based on those algorithms, building the models automatically. Compared with machine-learning AI, rule-based AI has advantages such as faster to automate tasks, a human can train AI, and lower cost. On the other hand, it also has disadvantages such as unable to learn independently, unable to train AI unless it is explicit knowledge, and unable to make decisions on matters it has not been trained on.

Machine learning can be divided into supervised learning, unsupervised learning, and reinforcement learning in general (**Figure 1**).
