5G, AI and IoT enable location management at ADR terminals

In an industry where safety, precision and speed are crucial, Van Moer Logistics was looking for a way to drastically improve the management of dangerous goods at their container terminal. As the operator of the largest tank container terminal in the Hamburg – Le Havre range, Van Moer Logistics manages an average of 1,250 container movements every day, spread over three terminal sites. On peak days, they process up to 1,350 trucks. Within this complex logistics ecosystem, it turned out to be difficult in practice to always know exactly where each tank container was located. This not only led to reduced efficiency, but also increased the risk of dangerous situations, especially given the chemical and flammable nature of the stored cargoes.

The European Seveso Directive requires terminal operators to be able to demonstrate the exact location of each tank of hazardous substances at all times. In reality, however, it turned out that at busy terminals between 10% and exceptionally even 60% of the containers were not in their correct location. After all, a container that is dropped off in the wrong place sets off a chain reaction in which other containers are also placed incorrectly, with the result that operators lose time searching or manually correcting. The impact of this extends beyond operational delay: in an ADR environment, any form of uncertainty equals potential danger.

To solve this problem, an innovative project was started in which four parties brought together their expertise. Van Moer Logistics took on the role of project leader and worked together with the Cronos Group – with its deep knowledge of Industry 4.0 applications –, telecom operator Proximus – responsible for the 5G network – and the University of Antwerp – which contributed its unique knowledge of innovative technologies to the port.

The common goal was ambitious, but clear: to develop a system that allows the location of each tank container at the terminal to be determined with real-time precision and this information to be continuously processed, secured and integrated with existing systems. That system had to be based on the latest technologies: 5G connectivity, artificial intelligence (AI) and the Internet of Things (IoT).

After a brief technical evaluation of different methods for determining container locations, we opted for a setup with a stereo camera that both captures images and measures the distance to each object. The recordings are processed in real time by a custom-developed AI model that first recognizes containers and then reads the unique container ID with OCR. In combination with the highly accurate GPS data from the accompanying antenna, we can continuously calculate where each container is in relation to the stacker – the vehicle that moves containers.

Because containers are often stacked on top of each other and therefore share the same GPS coordinates, an additional algorithm was built in for each video frame that determines the stack level . This means that we not only have latitude and longitude, but also the exact height position of each container.

A key challenge here is the huge data volume: to reduce data transmission costs and save the stacker battery, the AI model runs locally on an edge device (an NVIDIA Jetson) and we only send the processed results via 5G to our cloud platform. This required careful optimization of performance and power management.

For the realization, we bundled the expertise within Xplore Group: OnEdge provided the hardware integration, InfoFarm developed and trained the AI model, Cloudway built the scalable cloud platform and Unlock'd provided a multi-device front-end that gives operators and yard managers insight into the current position of each container at a glance. With all these competencies, we were able to turn the proof of concept into a completely working end-to-end solution.

What sets this project apart from previous attempts is the crucial role of 5G. The technology not only provides the bandwidth and reliability needed for real-time communication, but also enables highly granular, stable connectivity across the entire site through techniques such as Massive MIMO, beamforming and slicing. In combination with a private APN, a fully secure and isolated network was set up that is only accessible to the customer's devices.

Where traditional networks often struggle with large amounts of simultaneous connections or signal pollution on large industrial sites, 5G proved its worth as a robust and high-performance backbone for this application.

The results did not lie. Within three weeks, the system was able to correctly locate 70% of all containers, with an overall detection rate of 96%. In time, the system will support the monitoring of up to 40,000 containers per day. In operational terms, this meant an average time saving of 33 minutes per shift for the stacker operators. These efficiency gains translate directly into cost savings on labour, fuel and stacker rental.

The project proves that what was an ambitious vision of the future for years is achievable today thanks to the combination of 5G, AI and IoT. Although the technology was developed with the complex ADR context as a starting point, it can be used in a wide range of ways. Think of other container terminals, distribution centers, logistics hubs or even indoor warehouse environments and retail spaces. Wherever object location, real-time data insight and security have to go hand in hand, this solution offers new perspectives.

What started as a proof of concept on a single stacker has now grown into a scalable and reusable solution that fully harnesses the power of 5G and is ready for wider adoption inside and outside the port world.