The human algorithm: Why optimizing shipping routes isn’t just a mathematical exercise

The global shipping industry finds itself in turbulent waters. Rising fuel costs, mounting pressure to reduce CO₂ emissions and increasingly stringent environmental regulations are forcing the sector to reconsider how it’s operated for centuries. Traditional shipping routes are falling short in terms of both economic efficiency and environmental responsibility. With the sector currently accounting for 3% of global CO₂ emissions—a figure that could be as high as 10% by 2050 if current trends continue—the search for sustainable alternatives has never been more urgent.

A team of researchers led by IE University Professor David Gomez-Ullate and including Dr. Daniel Precioso Garcelán is tackling this challenge by applying advanced mathematical algorithms. They have developed the program Weather Navigation, which uses real-time weather data to optimize shipping routes. This technology could reduce fuel consumption and lower emissions by up to 10%, while also improving safety and arrival-in-port estimations. It offers a much-needed answer to a complex question—and arriving at this answer required more than just technical investigation.

A human approach to technological innovation

The point at which huge strides in technological advancement depend on more than academic insights; they also rely on the relationships between people, and the extent to which they’re prepared to trust, communicate and collaborate with one another.

Beyond its impressive technical achievements, perhaps the most fascinating aspect of Weather Navigation is the interdisciplinary, collaborative nature of the project. “To bring the project to life, we relied on the input of a large number of people from a variety of different fields,” Dr. Precioso explains. “The project wouldn’t be possible without their individual expertise.”

The internal Weather Navigation team is extensive. It includes experts in data science, physics, mathematics, data analytics and business management, computer science, computer engineering, and international relations and affairs. Together, these contributors represent several universities, including IE University, the University of Cádiz and the University of Cartagena in Spain, and Dalhousie University in Canada.

Bridging the gap between academia and industry 

Given the makeup of the research team behind the project, it should come as no surprise that Weather Navigation was first conceived as an academic idea. “At first, we simply wanted to know whether we could use weather patterns to optimize shipping routes,” Dr. Precioso comments. “We framed it in a very scientific and mathematical way. We didn’t see the ocean as the ocean, for example; we thought of it as a vector field. And we used concepts like gradient descent and variational optimization to frame a lot of our early thinking.”

Gradient descent is a mathematical technique that finds the best solution by repeatedly moving in the direction that most improves the outcome, like rolling a ball downhill to find the lowest point. Variational optimization refers to finding the best path by considering all possible variations and choosing the one that minimizes or maximizes a specific objective, such as fuel consumption or travel time. Applying these mathematical concepts helped the team to see real-life challenges with fresh eyes and fueled novel ideas on how to address them.

But these scientific approaches don’t necessarily work in practice. Or at least, they often need to be adapted in order to be understood, accepted and applied by industry players.

To bring the project to life, it was clear to Dr. Precioso and his colleagues that they would need to bring in experts from various fields. They soon broadened their team to include meteorologists, oceanographers, naval engineers and shipping industry professionals. They needed people who understood the sector’s challenges and requirements first-hand in order to bridge the gap between academic investigation and real-world application.

Speaking the same language

This gap became evident immediately in things as minor as the terminology used by the academic and shipping teams. As they started to work together, they noticed that they often had different words to refer to the same concepts.

This may seem relatively insignificant—surely it’s not all that difficult to exchange one word for another—but the reality is slightly more complex. Many studies have demonstrated how inconsistent terminology, alternative research methods and language barriers often get in the way of meaningful interdisciplinary research and development.

In a 2012 paper, University of Maine academic Susan Gardner elaborates: “Despite its growth, interdisciplinary research is often difficult to carry out, given the inherent challenges of bringing together multiple disciplines and, therefore, multiple ways of knowing and conducting research. In particular, disciplines have their own specific cultures, languages and standards for adequacy, as well as their own paradigmatic assumptions.”

Getting the terminology right matters; it’s how you get a larger team on board. As Dr. Precioso and his colleagues learned to navigate this jargon-filled landscape and grasp the language used by their industry peers, the relationships between everyone involved became stronger.

Navigating priorities for real-world application

Understanding the different perspectives that each team brought to the project became essential in setting goals that worked for all parties. For instance, Dr. Precioso noticed that the mathematicians tended to have different priorities than their industry counterparts. “They were primarily focused on optimizing the trajectory of a ship in open water, because this scenario means that we’ll always be able to find the best possible route,” he explains. “But we knew that the engineers and shipping company experts we were consulting with would be more interested in the route’s operational functionality.”

While it was tempting to focus only on open-water sailing since this would deliver the best results in terms of fuel optimization, in practice, this simply isn’t viable. Open-water sailing isn’t the only scenario ships encounter in real life. They also have to maneuver around natural obstacles like coastlines, islands, archipelagos, coral reefs and even icebergs.

As climate change propels the melting of the polar ice caps, new shipping routes are emerging in the Arctic. Since ships are traversing through ice in these environments, routing algorithms need to be able to predict the location of icebergs and where they’re likely to move to, so that ships can navigate around them safely. Ships also encounter maritime hazards like other vessels, oil rigs, floating debris, restricted waters and zones that have been established to protect marine diversity. Weather-related obstacles like storm systems, fog banks and rough seas need to be taken into account, too.

In addressing this challenge, the teams ultimately developed a two-stage pipeline. First, they created a stochastic algorithm, a computational method that uses randomness and probability to solve problems. This means the tool can produce different results every time it runs, even with the same input data. “This approach unsettled our mathematical team, but it helped us to account for obstacles and to ensure that our hypothetical ships wouldn’t hit them,” Dr. Precioso explains.

Once the mathematicians knew the ships could travel safely from A to B, they adjusted their original metrics to optimize the route. This made the recommended route not only safe and viable but fast and effective, too.

“The more constraints you add, the more difficult it is going to be for the algorithm to find an optimal route,” Dr. Precioso says. “But we don’t have a choice. We have to overcome these requirements if our tool is going to be useful and if we’re going to see widespread industry adoption.”

Building trust with industry partners

Adoption is always a challenge for any new technology. The reasons may be economic, infrastructural, educational or something else entirely. Whatever the case, convincing a company or sector to deviate from past practices is rarely a smooth transition.

“I think the biggest adoption-related challenge for us is building trust,” explains Dr. Precioso. “We have to show that our tool is reliable. And that’s not a simple task because shipping companies are unlikely to put their crew and cargo at risk for a novel piece of technology they haven’t tested yet.”

Dr. Precioso and his team have also found it difficult to validate their results. This is largely because such an exercise would need to involve two ships setting sail in the same conditions at the same time—one taking the standard route and one taking the optimized route. Since this can’t be done realistically, the team is forced to rely on estimations.

However, by partnering with shipping companies, the team has been able to strengthen their predictions. “We’ve been working closely with Boluda Corporación Marítima, a leading company in global maritime services, to demonstrate the accuracy of our work,” Dr. Precioso adds. “They provided us with the details of previous routes so that we can estimate their fuel consumption. Once they confirmed that our estimates were correct, the journey to building trust began.”

The next step involved demonstrating the alternatives Boluda might have considered had they used the Weather Navigation tool, and how effective these amendments would have been in terms of fuel consumption, CO₂ emissions, time at sea and arrival at port. “Working with previous trips and showing how they could have been improved has been crucial,” he continues. “With every new metric addressed, we’ve found that our potential clients have shown further interest in our work and its application.”

Taking feedback on board has been another essential component. Weather Navigation’s computing teams have ensured that the algorithm can leverage ocean currents, wind patterns and wave data in real time, so that its recommendations are always up to date. “Our collaboration with ship captains and the owners of shipping companies has reiterated the importance of this data being both instant and accurate. We have taken these requests seriously. And I’m grateful that we are working with incredibly competent computer engineers to make them possible.”

The humanity at the heart of technology

After years of highly technical research that involves some of the latest advancements in a multitude of fields, the success of this innovative technology comes down to several very human factors: language, understanding, trust, connection and collaboration.

“There are so many different factors at play when you launch a new piece of technology,” Dr. Precioso concludes. “The different experts involved each come with different objectives, priorities and perspectives, and this often changes the nature of the project, presenting quite human challenges in the midst of everything else. Ultimately, so much of this sort of work is defined by human relationships.”

This is not to negate the value of the technology itself, or to undermine the impact that Weather Navigation is likely to have on the shipping industry’s pursuit of effective, sustainable interventions. But we can’t forget that even the most sophisticated solutions require something fundamentally simple: people willing to listen, learn and trust one another across the boundaries of their expertise.

For now, the tool is still being developed, though users can already log in to the online portal and insert the route they want to travel. In just moments, the algorithm produces a suggested route, informing users of the weather conditions they’re likely to encounter, the velocity they should travel at and how much fuel they’re likely to save.

IE University’s researchers and their extended interdisciplinary teams are working towards the real-life adoption of this project by shipping leaders. Weather Navigation is poised to provide an essential piece of the puzzle for an industry seeking sustainable solutions.