"Delays, bottlenecks and CO2 emissions" – Placing the proverbial finger on the wound, Adriaan Landman began the discussion by unpacking the three-pronged challenge confronted by port and terminal operators in the first seconds of the show. The COO and co-founder of the tech start-up AllRead in Barcelona immediately provided an explanation for the problem: “More than 95 per cent” of the players have “no automated solutions to control incoming and outgoing containers and vehicles”. Here, AI-based technology for automatic optical character recognition (OCR) offers new opportunities to optimise processes and increase efficiency.
Further interrogating this point, an audience member questioned the what the potential value-add from AI specifically could be, considering the existing mature state of gate control systems in large container terminals. These systems typically deploy traditional number plate or waggon number recognition and damage control using surveillance cameras. Adriaan Landman explained that a Deep Learning-based OCR solution relieves from "technological and financial limitations" of existing systems because it reduces hardware dependency. ‘AI can put any camera in the right place, process images in real-time and deliver results to the terminal operating system’. "This makes access control much more cost-effective for ports of all sizes," he says, citing a key argument for its use in ports.
"It's about increasing throughput, lowering operating costs, reducing CO2 emissions, but also increasing safety," says Stephan Piworus, Vice President Europe at the start-up atai, headquartered in Hyderabad, India, summarising the advantages of AI for maritime operations. ‘So far, this has been held back by a port environment with many different systems "that do not interact well with some operators". Instead, a holistic approach from gate to rail to warehouses is essential’.
The concerned port industry however is not so easily convinced by AI. As a graduate in business administration with a wealth of terminal operator experience, Stephan Piworus understands the collision between "a conservative industry" and new technologies: "People don't see the potential, but rather the risk at first.” This ranges from mistrust regarding the service level of start-ups, which lack success stories in the industry, to a fear of a complete upheaval in the port sector’. Stephan Piworus, however, does not think a company’s tenure in the industry is crucial: "It's more about the concept and how easy it is to maintain and operate."
Start-ups that do not originate from the maritime industry seem to bring especially new perspectives and agile problem-solving approaches. At AllRead, according to Adriaan Landman, not only is the permanent learning curve of the developers increasing, but the algorithms are also constantly learning. Once installed, AI solutions can thus be improved technically and adapted to new rules or processes that need to be integrated into the operation.
Responding to a question in regard to the ideation to prototyping time frame of startups, Alois Krtil, CEO at the Artificial Intelligence Center Hamburg (ARIC) surprisingly asserts, "In my experience, this phase often doesn't end at all". Yes, it is a momentous challenge to be continuously stuck in a prototyping process. Further responding, the business information specialist explains, if start-ups "are currently working on generative AI, for example, i. e. new paradigms that are really fast", this can reduce development times by up to 50 per cent. At ARIC, there is "the first product version in about four to six months". Stephan Piworus further commented critically that: A fast prototype from the laboratory environment often does not work under terminal conditions, for example, when sunrises and sunsets limit camera functions. Therefore, he considers it "very important to understand the business processes".
Questions from the audience poured in to the panel via chat: ‘How can efficiency and cost be improved with AI? Alois Krtil estimates cost savings of "up to 50 per cent" in B2B environments "based on everything we've learned in various sectors over the past few years". Objection from the moderator, whether the port industry with its challenges is not unique? Alois Krtil explains that when it comes to basic principles such as data analysis, real-time capability of systems or energy- and cost-efficient hardware developments, there are "of course overlaps with other industries". For example, applications that are also used in smart cities for traffic or critical infrastructure are suitable for monitoring in the port.
The next sticking point: How secure are AI solutions, also regarding compliance with relevant regulations? After all, ports operate with a lot of sensitive data on ship movements, cargo information or customs. Adriaan Landman explains that AllRead meets all national cybersecurity requirements in Spain but is constantly on guard "because customers sometimes have even higher expectations and may anticipate new regulations". On the discussion of whether data protection laws hinder innovation, Alois Krtil has a clear opinion: "We need more progressive regulation." The EU Data Act points in the right direction but needs to be "implemented in a really pragmatic way".
Final question in the round, what might the future of port AI look like? While Alois Krtil observes a "strong trend towards trust-based AI", Stephan Piworus encourages starting early with AI and preparing employees "so that fear does not overcome curiosity". Ultimately, AI is only a "means to an end", says Adriaan Landman. The goal is to solve problems - whether with AI, digital twins, or automation.
The complete session Connecting Ports #04 is available as a video recording here.
