There are currently over 20 million shipping containers globally, and six million of them are traveling around the world on vessels, trucks, and trains. In total, they make about 200 million trips a year. Containers have an identification number and type indication relevant to the logistic operations at container terminals. This includes seaports and inland terminals, the smaller ports in the hinterland where containers arrive via barges, trains, and trucks. As such, “container recognition is relevant at various stages of the terminal operation,” indicates Sander Maas.
The automatic detection of containers proposed by Sentors has several benefits. For example, the solution brings a relevant time-saving aspect, as the truck drivers and terminal planners avoid wasting time by manually writing and validating the numbers for identifying the cargos. It also facilitates all kinds of insurance mechanisms to settle damage disputes, as it is essential to have visual proof of a container’s condition before entering the premises.
Seaports, such as Rotterdam, have been using automatic recognition systems for years. All incoming and outgoing trucks pass a gantry equipped with various cameras and lighting units. However, inland terminals have a much smaller volume and do not have a business case to justify a costly investment. For this reason, very few inland terminals have employed automated systems, and most of their operations involve manual actions to enter numbers and take photos when damage is spotted.
As Sander explains, “Sentors has developed a 4G-based solution for inland that only needs power and a fence or pole to mount to. In order to obtain robust results on all sorts of containers and in adverse conditions, the solution is leveraging on recent developments in machine learning. The result is a highly reliable, flexible, and low-cost solution that brings crucial benefits to a much larger market than only the seaports.”
Machine learning models are a game-changer in terms of robustness and flexibility. However, they face one practical drawback: they are computationally expensive as machine learning engines processing streaming video requires a Graphical Processing Unit – GPU. For this reason, Sentors has chosen for edge processing, that is, to use an outdoor-grade hardware system that comes with a GPU and 4G modem placed next to the camera.
Although the installation is straightforward, it still requires hardware and local software maintenance. The less needed, the easier it is to deploy such solutions in harsh environments such as cranes and reach stackers (the large “forklifts”) that lift the containers.
Here a key interest in 5G comes into play: if the machine learning software can run elsewhere, such as somewhere in the telecom network, it would enable an entirely new way of rolling out such solutions. That would result in lower system integration costs and new options for installations in harsh environments. Next to that, it also opens opportunities to use mobile phones for this purpose and uncovers new possibilities when combined with Augmented Reality.
The merits of 5G’s high bandwidth and low latency become apparent when working on video streams where there is a need to render a real-time overlay on the screen. Like in many other areas, there are several compelling applications for Augmented Reality in the domain of shipping containers.
For example, train drivers are required to check each container according to a planned list visually. Since this revision may involve 30 or more containers that all have a very similar number, mistakes do occur – not to mention the time and energy invested in the process. Sentor’s current prototype is exploring a new solution that uses the power of smartphones. By utilizing the convenience of a smartphone with 5G, users will be able to employ the system from anywhere. Furthermore, by integrating 5G, a stable connection can be guaranteed at all times. This way, users are no longer restricted to a bound location where the hardware is situated. The smartphone streams the video to a (local) server, and the server, in turn, processes the stream and sends back any important information that is relevant to the user.
Additionally, the implementation process of the solution has been generalized and streamlined so that it can be deployed on various systems. Another significant advantage is that it eliminates the need for specific required hardware, giving the companies more freedom in their budget. These new features are being designed and developed at the 5G Hub by Tim van Oudheusden, under the supervision of Edwin Dijkstra.
Sentors contacted the 5G Hub and started a project together to explore the anticipated benefits of using 5G over existing possibilities. As Sander says, “we are pleased to get access to the facilities and competence the 5G Hub brings, allowing us to further develop our product portfolio. For our solutions, 5G holds the promise of easier maintenance, new installation possibilities, and Augmented Reality applications where the human inspectors are offloaded.”
Sander Maas talks about the collaboration with the 5G Hub with enthusiasm: “We are excited to work with the 5G Hub. In particular, our current demo-set-up at the 5G Hub shows the aforementioned Augmented Reality application.”
According to Sander, with the help of Ericsson and VodafoneZiggo engineers and the students who worked on the project, Sentors is well-positioned to make full use of the capabilities offered at the 5G Hub. “Our project is not limited to shipping containers. We also have solutions for several different applications such as outdoor stock counting. Every new step we take in the 5G Hub can instantly be applied to these other applications as well,” he says.