Deep Learning

Artificial intelligence (AI) that helps to check containers for damage – that is the goal of the COOKIE research project. In the long term, it would shorten the process and customers could regain access to empty containers more quickly. An important contribution to “just-in-time” logistics in the maritime industry.

At HHLA’s subsidiary HCCR, thousands of empty containers stack up waiting to be thoroughly checked – still manually – for damage and might also require to be cleaned or repaired. Only after that can they be loaded onto a container ship. In the future, we are set to see more targeted classification and processing of damage made possible through the use of artificial intelligence.

The first step is developing an AI-supported application for image recognition that can identify and assess damage. Our project partner, the Fraunhofer Center for Maritime Logistics, is currently working on this. With the help of deep learning, AI is learning using approximately 1.8 million classified images of damage. In future, inspectors will be supported by automation while identifying and assessing damage to empty containers.

Digital container registration

It is the very definition of a Sisyphean task: Upon receiving its test certificate, an empty container leaves the yard of Hamburger Container- und Chassis-Reparatur-Gesellschaft (HCCR). But a new sheet steel box is already waiting to take its spot. As long as operations continue, there is no end in sight.

Most of the empty containers stored in the Port of Hamburg must be inspected for damage and contamination by the service company HCCR, so that they can then be professionally cleaned or repaired where such is required. Only after this process can they once again be loaded onto a container ship in accordance with international standards (CSC / UCIRC).
Inspectors, or “checkers” in port jargon, examine every single one of what amounts to thousands of containers every year, inside and out. Are there nails protruding from the floor panel? Are the walls rusted through or dented? Can the required stability still be guaranteed?

Specialist HCCR workers photograph every noticeable spot using a modern, hand-held industrial device (similar to a mobile phone) and enter standardised damage codes into software that has been specially developed for the repair of containers. These are then sent to a database via Wi-Fi, which automatically calculates the estimated costs for the pending repairs. The repair order is then sent to the hand-held device of a steelworker. After the work has been carried out, he documents the proper completion of the repairs by taking photos.

Deep learning fights the lack of containers

This process costs time and thus money. In addition, there is a global shortage of steel boxes. “Our customers desire significantly more qualified inspections of their containers than we are able to provide, despite digitalisation,” says HCCR Head of Sales, Toni Jakat. “The percentage of containers which are found to be undamaged during checks is a high double-digit figure. If we were able to identify these intact containers from the outset and the checkers were able to concentrate on those which are actually in need of inspection, we could achieve a significantly greater level of efficiency in fulfilling our customers’ requirements.”

Separating the wheat from the chaff, or in other words, ensuring that the detection of damage is as accurate and comprehensive as possible, is the primary goal of COOKIE, a digitalisation project that is scheduled to run for a period of 30 months. The acronym stands for “COntainerdienstleistungen Optimiert durch Künstliche IntelligEnz”(AI). In English: Container Services Optimised Through Artificial Intelligence. Toni Jakat launched the project and brought together the various partners involved. He submitted a funding application to Germany’s Federal Ministry of Transport and Digital Infrastructure (BMVI), which was subsequently approved, awarding 900,000 euros of funding as part of the “Innovative Technologies in Ports” programme.

Rethink. Think further. Think ahead.

Megatrends and our new business areas Automation, Artificial intelligence, Interconnected supply chains, Drone technology and Hydrogen

Discover innovation at HHLA

Image recognition processes with Deep Learning.

At HHLA, HCCR’s parent Group, there are a total of nine such IHATEC projects underway, which emphasises the logistic company’s importance as a technological leader in the field of digitalisation. The project partner is the Fraunhofer Center for Maritime Logistics in Hamburg-Harburg. It is developing an adaptive algorithm for image recognition processes, that is to say, a form of artificial intelligence (AI). It shall be capable of detecting the current status of a container and providing a reliable evaluation of such. In order to do this, it must first analyse thousands upon thousands of stored photos. This process is known as deep learning.

“The AI shall compare a live image against stored images of damage,” states Jakat, summarising the aim. “This will mean that we will no longer have to deal with boxes that are suitable for shipping, but can instead tackle a higher number of repairs.” Having fixed cameras automatically photograph every square centimetre of an empty container as it passes through a heavy goods vehicle checkpoint at the HHLA terminals would be a great research success. This would allow the percentage of undamaged containers to be separated from the rest.

Another aim of COOKIE is sustainability as, in a very similar way, the AI may one day be able to help HCCR use its tank container washing facilities more efficiently. Huge quantities of water and chemical additives are currently used to clean persistent contamination inside the tanks. According to Jakat, COOKIE opens up exciting possibilities for the washing facilities too: “If the IT system is able to use automatic algorithms to correctly evaluate how severe the contamination is, then the wash programme for each tank could be individually optimised.” This would help to reduce the consumption of resources.

More innovation with artificial intelligence at HHLA

Contact persons