AI Systems for Virtual Shielding of Logistics Tunnel

Industry 4.0 calls for Logistics 4.0: the digitalisation of order management and goods transport processes is now an unstoppable trend for large enterprises.

This process, which has become even more urgent since Covid19 and the consequent rapid increase in online sales, is supported by the continuous evolution and integration of technologies such as Artificial Intelligence and the Internet of Things, introducing radical transformations in terms of Smart Logistics.

AI: Artificial Intelligence What it is and What it does

Artificial Intelligence (AI), as defined by the European Commission in 2018, is “the ability of machines to display human capacities such as reasoning, learning, planning, and creativity.”

Thanks to these characteristics, Artificial Intelligence – a generic term that includes many different types of technologies and applications, as we will see – is able to understand its environment, process what it perceives, and identify solutions for specific objectives: in other words, the system receives data previously prepared or collected with sensors, processes it, and ultimately provides the requested answer.

If we look at the definition provided by the Osservatorio Artificial Intelligence of the Politecnico di Milano, AI is “the branch of computer science that studies the development of Hardware and Software systems equipped with specific capacities typical of human beings (interaction with the environment, learning and adaptation, reasoning and planning), able to autonomously pursue a defined objective, making decisions which had previously only been entrusted to people.”

The scientific community has defined two types of artificial intelligence based on certain parameters: the capacity of these technologies to imitate human characteristics, the technology they use to do so, and their applications in the real world.

We can therefore speak of:

1. Weak Artificial Intelligence, which includes systems able to simulate certain human cognitive functions, but which neither imitate nor replicate human intelligence. This type of AI is geared towards objectives and designed to carry out individual tasks;
2. Strong Artificial Intelligence, which includes systems capable of becoming knowledgeable, that is, of imitating human intelligence and/or behaviour.

The Artificial Intelligence technologies used today fall under the first category, but achieving strong artificial intelligence is a challenge no longer relegated merely to the world of science fiction.     

It was 1956 when a conference at Dartmouth College in Hanover, New Hampshire, marked the birth of the ‘Artificial Intelligence’ discipline, followed by decades of intense experimentation and alternating periods of stasis and progress, until now, with AI having become a primary field of interest for computer science experts, but also the technology which more than any other has defined the last decade, set to also influence the next.

And that’s not all: the ethical implications and social impact of AI are now such that in 2021, even the European Commission presented a proposal for a Regulatory framework on Artificial Intelligence to regulate the development, use, and marketing of these technologies.

What is Artificial Intelligence used for?

AI systems are used in many different ways in today’s society and economy, with certain applications exploiting this technology without us even knowing it.

For example, to name a few:

• in e-commerce and online shopping, AI can provide suggestions based on previous purchases, searches, or other behaviour adopted on the web, thus influencing the customer journey;
• many of the devices we use every day have virtual assistants which give suggestions, find solutions, or respond to our questions;
• in our homes, AI-based systems are used to optimise consumption, in cities, artificial intelligence can help improve mobility and reduce traffic jams;
• automatic translation software, or those which add automatic subtitles to videos, use AI to improve the results;
• and finally, even driverless vehicles, which are no longer just science fiction.

Artist: KOTA YAMAJI

What are the types of artificial intelligence?

According to the Observatory on Artificial Intelligence, there are 8 types of Artificial Intelligence applications, distinguished according to their purpose of use:

1. Intelligent Data Processing (IDP): algorithms that analyse specific data to extract information and perform consequent actions (for example, predictive analysis);
2. Virtual assistant/ chatbot: software able to perform actions or provide services for an individual based on commands received either vocally or as text (systems commonly used in company Customer Care);
3. Recommendation system: applications that direct users’ choices based on information they have provided (for example, systems that suggest a purchase based on buying habits);
4. Natural Language Processing (NLP): solutions that process language (for various purposes: understanding content, translation, production of texts);
5. Computer vision: studies algorithms and techniques to allow computers to reach a high level of understanding of the contents of images or videos (systems widely used in video-surveillance);
6. Autonomous Vehicle: the first of the three physical AI applications, consisting in self-driven vehicles;
7. Intelligent Object: objects able to perform actions without human intervention, and make decisions based on the surrounding environmental conditions;
8. Autonomous Robot: robot able to move without human intervention, based on information collected from the surrounding environment.

The AI market in Italy

The Osservatorio Artificial Intelligence of the School of Management of the Politecnico di Milano recently published the results of its study on the artificial intelligence market in Italy.

This is a market which grew by 27% in 2021 compared to the previous year (when it was affected by Covid19), worth a total of 380 million euros, a value which doubled in the last two years, represented by Italian projects for 76%, and exported projects for the remaining 24%.

In the Italian AI market, in 2021, the podium was taken by Intelligent Data Processing (35%), Natural Language Processing (17.5%), and Recommendation Systems (16%). These were followed by Chatbox and Virtual Assistants (10.5%), Computer Vision initiatives (11%), and Intelligent Robotic Process Automation solutions (10%).

In a context where all types of artificial intelligence applications are rapidly growing, the areas which experienced a more marked acceleration with respect to 2020 are the Computer Vision (+41%), Chatbot and Virtual Assistant (+34%) and Intelligent Data Processing (+32%) fields.

As already observed in the industrial IoT, a significant gap is also emerging in AI between large corporations and SMEs, with 59% of the former having launched at least one AI project in 2021, compared to 6% of SMEs (with only 2% representing fully implemented projects).

The data relative to consumers is just as interesting, insofar as almost all (95%) admit to having heard of AI, a figure which drops to 60% if we analyse the ability to recognise the presence of AI functions in the products/services used.

In line with the European Strategy, in 2021, even Italy adopted a ‘Strategic Program for Artificial Intelligence (AI) 2022-2024’, outlining twenty-four policies to implement during the 3-year period between 2022-2024 to enhance the Italian Artificial Intelligence system.

Logistics management can also be improved by computer vision, which as we have seen, allows computers to reach a high level of understanding of the contents of images or videos: this technology can be used to systematically unload goods, to identify damaged raw materials and products, and even locate items and packages in the warehouse or point of sale.

Even Intelligent data processing, and in particular predictive analysis can be usefully applied to logistics: suffice to think of how an analysis of historical data can optimise the shipping and delivery process thanks to “intelligent” predictions of consumer behaviour.

Last but not least, the ability of AI to manage big data, that is, enormous amounts of data derived from an increasingly digitalised logistics process, in a highly sophisticated manner: thanks to artificial intelligence, it is essentially possible to evaluate data from multiple sources (robots, vehicles, smart devices) in real time, and use it for analysis, with a significant impact on the efficiency of the entire logistics chain.

As highlighted by the study conducted by the Contract Logistics “Gino Marchet” Observatory of the School of Management of the Politecnico di Milano, regarding the Contract Logistics market in Italy, (which, with +3% compared to 2020, recorded a turnover of 86 billion Euro in 2021), logistics is also embarking the path of a sustainable transformation, which in the case of warehousing activities, also affects the handling and automation processes of the warehouses themselves.

Logistics 4.0: Package handling and RFID

For companies making their way in the world of Industry 4.0, Logistics 4.0 entails new storage, handling, and transport systems developed along the lines of physical automation, connection, and the decision-making process.

Important transformations have already been introduced in terms of Smart Logistics by the revolution of the Internet of Things, which, by applying a sensor system and Rfid readers to objects, allows ‘communication’ between resources, which incredibly facilitates package handling, storage, and transport systems: for example, by using Rfid tags on the trolleys used to handle goods, the picked packages can be automatically read.

Industry 4.0 is moving towards solutions increasingly aimed at integrating the Internet of Things and Artificial Intelligence, two technologies which when combined, can lead to radical transformations: suffice to think of the potential of a system able to modulate the production program of a product based on the incoming orders, or pieces in transit on the line.

The project is structured across three levels:

1. Redesign of tunnel HW: making use of the company’s internal and external experts, we have redesigned our tunnel, developing a new shielding with smaller dimensions, modifying the position of the antennas and selecting a new RAIN RFID reader and new RAIN RFID antennas.

2. Redesign of data architecture: we have improved our architecture in an IoT perspective, simplifying the system’s implementation and increasing its security.

3. Artificial Intelligence System to process the readings (described below).

The idea stems from the desire to simplify the tunnel structure, and at the same time improve its reliability, adding an artificial intelligence engine capable of correctly classifying the tags that belong to the package transiting through the tunnel, excluding any external, and therefore unwanted readings.

The first studies began in July 2019, but the project truly began to take shape in the summer of 2020. After the initial attempts, we realised the objective was achievable, but needed the support of a partner with specific skills in order to reach the set goal, which was 99.9% accuracy on the items. 

After issuing a call for tender among certain companies, we decided to develop the project with Data Life, which had already partnered us for several years for t!Insight. Data Life has now completed a first stage of in-depth analysis, confirming our initial internal results, demonstrating that the path chosen could effectively produce the desired results.

The AI engine able to correctly discriminate the tunnel readings allows us to obtain the following results:

    • Fewer errors in logistics.

    • Gates with more contained shielding, which are therefore less costly and less cluttered, allowing installation in logistics points that were previously inaccessible.

    • Simpler implementation of the UHF.

    • Fewer restrictions on use of the tunnel, such as being careful not to overstep the current reading areas.