Behind Every Delivered Package Lies Precision-Tuned Air Logistics

As Europe’s leading hub for air-based e-commerce, Liège Airport is relying on new technologies to boost its efficiency. Its main challenge? Loading and unloading cargo planes.
Its solutions? AI, IoT, and digital twins. Enriched data to manage day-to-day operations—and predict the future.

By: Michaël Schyns and Jenny Tonka

Ordered something online for the holidays? There’s a good chance your package passed through Liège Airport (LA) in Belgium. Thanks to its strategic location at the heart of Europe, the airport now processes 3,000 tons of freight every day and serves nearly 400 million consumers. This makes it the #1 full-cargo airport in Europe, #5 among combined passenger-freight airports, and places it in the global top 20.

Need your deliveries to arrive fast? Liège Airport is ideally located—within a day's drive from major European cities, close to an autonomous port and key rail infrastructure. But the sheer volume of goods being handled makes the loading and unloading of cargo aircraft a true logistical challenge.
Air freight ground handling—specifically the coordination of service vehicles—is considered one of the most critical parts of the supply chain. A single failure at this stage can cause significant delays, with major economic and environmental consequences.

In short, ensuring your package arrives on time means this process must be optimized and strictly controlled. Easier said than done. The process involves many actors and is subject to a wide range of strict constraints. On top of that, Liège Airport’s zero-carbon strategy further limits the available options. While traditional operational research methods have proven useful, they’re now hitting their limits. New approaches and technologies are needed—AI, digital twins, even virtual reality.

We studied this issue as part of the European Interreg project “Digital Twin Academy” and in a PhD project funded by the FNRS.

30,000 Aircraft Movements a Year
The challenge lies in coordinating all ground activities and determining the route of each vehicle involved in aircraft services—fueling, cleaning, cargo transfers, etc.
This is a well-known operations research problem: the Vehicle Routing Problem (VRP). However, due to its complexity, research usually focuses on simplified cases involving a single type of vehicle.

Airports, by contrast, are bustling ecosystems where multiple operators work in concert, using many different types of service vehicles. Liège Airport handles over 30,000 aircraft movements annually. At the height of the COVID-19 pandemic, it was designated by the WHO as the European hub for distributing medical supplies. The airport also operates a vast fleet: over 1,500 container dollies (ULDs), around 100 towing tractors, 50 high-loaders, and about 30 different types of trucks.

Tight Synchronization Constraints

Liège Airport faces complex cascading synchronization challenges across time and space between vehicles.
For example, before a tractor can take a container to the warehouse, it must receive it from a speedloader, which must first collect it from the high-loader operating at the aircraft. These routes cannot be planned independently.

Data, Data, and More Data

To address this, we developed a synchronization-based optimization algorithm. It calculates the best routes for service vehicles to minimize the total time needed to service each aircraft. Tested on real historical data from Liège Airport, the results were conclusive: with the right technologies and approaches, delays can be reduced and fewer resources are needed.

However, this only solves part of the puzzle. Another major challenge is data access—essential to feed the algorithm and move from theory to practice, from virtual to real.
Real-time ground handling control requires precise tracking of all vehicles and services in progress. Some data, such as flight schedules, are available through airport information systems.
But due to the number of operators and the diversity of often partially confidential, incompatible IT systems, no airport has a fully integrated system covering all activities. Fixing this has become a top priority.

5G, IoT, and AI

Data needs raise new questions about technology choices. In our research—and in the 5G era—our first approach was to use the Internet of Things (IoT), equipping service vehicles with GPS sensors.
This method revealed limitations: signal loss indoors, unreliable mounting, and limited battery life. The GPS accuracy also proved insufficient, prompting us to explore other technologies.

We are now focusing on computer vision-based artificial intelligence. Since airports are typically equipped with extensive camera networks, we decided to leverage them. AI analyzes images in real time to automatically detect vehicles operating on the tarmac. A tracking tool then monitors aircraft service progress at any given moment.

Liège Airport and Its Digital Twin

Now that we have an algorithm predicting what operation should be happening at any time—and real-world data showing what’s actually happening—we can move on to monitoring and control. Enter the concept of the digital twin. The algorithm acts as a virtual replica of the airport, updated in real time through a constant two-way data stream with the physical world.
The digital twin is fed by reality—and in return, provides feedback that the real world can use. A continuous feedback loop is born.

As a cherry on top, a 3D model of Liège Airport has been developed to visualize this loop. With a VR headset, you can immerse yourself in this digital twin as if it were real—minus the risks. This virtual model can simulate scenarios, manipulate time, or train operators in near-real conditions.

As you can see, getting your package delivered on time is much more complex than it appears.

With the initial contribution from Arnaud Stiepen, expert in science communication.

Translated from the original French version published in The Conversation.