Promising fields of application for AI in Port of Hamburg

Congestion involving trucks in the port costs time, money and nerves and is often avoidable. Traffic flows in the port can be improved using artificial neural networks to predict the number and time of arriving lorries. The Fraunhofer CML has developed a model that uses a digital image of the handling processes of logistical nodes for improved handling.

"Our algorithms learn from past data what kind of arrival rates and handling times can be expected in which situation, and can determine peak times in advance. The forwarders can then react to this information to avoid waiting times. The transparency created for all those involved creates a self-regulating system."
 

Lorries have been dispatched in time slots to avoid congestion in recent years. Kretschmann remarked, adding: "That is a clever solution but it is inflexible and can have disadvantages. The better the database, the more accurate the forecast. "If we can integrate the lorries' telematics units in future, that would certainly help achieve more accurate forecasts," he noted. The AI-based forecasting approach can be applied to many other use cases such as ship calls or to transport empty containers.

The provision of empty containers in global supply chains comes to an estimated USD 20 billion p.a. The "Container Availability Index Made in Germany"  (C-TIMING) aims to reduce empty runs through innovative forecasting methods that relieve both budgets and the environment. "No transport can take place without an empty container at the loading site. Optimal distribution of loading units worldwide can be done by each shipowner or they can join forces. Our co-operating partner, xChange Solutions GmbH in Hamburg, has developed a platform to bring users and owners together," said Kretschmann.

The Fraunhofer CML now wants to take C-TIMING project a step further. "Our goal is to calculate where, when and how much demand can be expected based on the latest research results using artificial intelligence." Millions of individual container journeys are evaluated, statistical methods and machine learning techniques are applied resulting in a model that predicts supply and demand in an index. "This information can be used for transport decisions and forms the basis of dynamic pricing. It takes the latest market developments into consideration when calculating the cost of container transport."

Empty containers are stored and inspected in depots until their next use. However, such checks are frequently unnecessary, said Kretschmann, adding: "It's called 'banging' in the port. Considerable force is needed to move the containers, which can be damaged swiftly. Identifying them reliably is hugely important for safe transport. Our goal is to find the damage using computer vision or AI-supported image recognition." To this end, the Fraunhofer CML is working with HCCR Hamburger Container- und Chassis-Reparatur GmbH on software for image-based damage detection as part of the COOKIE project. "Image recognition has made enormous progress in recent years and can now be integrated into many practical applications in logistics."

Kretschmann expressed confidence that AI will gain even more ground in the maritime sector. "Thanks to increasing amounts of data for AI applications, more and more solutions can be found. At the same time, data storage and transmission have become much cheaper and the development of smart sensors is progressing." And companies in the maritime sector are showing keener interest. "The prospects for the future are good," Kretschmann stressed. "Similar to the retail and banking sector, the maritime sector offers ideal prerequisites as plenty of data is generated in transport."

ys/kk/pb