Swarm intelligence
ETH students on a project-based programme are developing underwater robots that can think and act collectively. It’s the perfect opportunity to see how research and technology can dovetail with teamwork.
“I’ll never forget the moment those three robots first swam together in the pool – they moved in perfect formation, just as we’d hoped,” recalls Alex Sheridan. As project manager for Swarm – one of 12 focus projects run by ETH Zurich’s Department of Mechanical and Process Engineering in the Spring and Autumn Semesters of the 2024/25 academic year – he fronted a group developing underwater robots that operate as a team rather than individually. “Controlling a single robot is fairly straightforward, but things get really interesting when you have multiple robots interacting,” he says.
ETH professor Melanie Zeilinger is well acquainted with such challenges. Her research explores how to control complex systems safely and reliably – even when they’re subject to uncertainty, changing conditions or human interaction. The Swarm project provided an ideal opportunity to put her work on self-learning, data-driven control systems to the test.
For the students, the key problem was how to get multiple autonomous systems to coordinate with each other in locations where sensor and communications systems are hampered. “Radio waves don’t travel very far underwater, which makes it a particularly challenging environment,” says Zeilinger. Underwater conditions are also in constant flux, meaning the systems have to contend with currents, signal loss and unexpected obstacles. “The students learned a huge amount – not just about solving technical problems, but also about putting theory into practice,” Zeilinger says.
As part of Aris, an ETH student initiative for hands-on aerospace projects, Swarm is one of a number of projects that show how robotics can bene?fit wider society. Robot prototypes developed in the lab might one day be used to monitor glaciers, study ecosystems or assist in disaster relief operations. Zeilinger emphasises that such projects create a useful bridge between research and society, helping us better understand and protect natural habitats. The three Swarm robots, for instance, could help map large underwater areas, detect and measure pollution and inspect submerged structures such as bridge piers.
From plan to prototype
The focus projects have been a cornerstone of the department’s Bachelor’s programme for more than two decades. Conducted during the fifth and sixth semesters, they are among the most popular teaching modules, giving students a chance to translate theory into practice. Teams of five to ten students spend the two semesters developing a prototype, moving from the initial concept through design and engineering to a final presentation. Here the students take charge of everything themselves, from conception and planning to development and production – plus budgeting and communications.
“In the first two years of their Bachelor’s programme, students complete all their compulsory courses. By the third year, they have already built the foundations they need to tackle a focus project,” says Alexander Hernández Oendra, educational developer at the department.
Zeilinger affirms the value of this hands-on learning. “It’s only when you develop a working system that you truly grasp how sensors, mechanics, control systems and software all mesh together,” she says. “For many students, this is their first real encounter with research and development.”
Sheridan vividly recalls the intense early phase when the team had to gel together and define a shared mission. It quickly became clear which ideas were overly ambitious. The team learned to triage tasks, break work into manageable chunks and align their vision with the constraints of available resources. That academic year yielded not only a working prototype but also a collective experience that none of them will forget.
Robots in action
This text appeared in the 25/04 issue of the ETH magazine Globe.
Engineering mindset
In Swarm, the team concentrated on the core challenges of underwater communications, localisation and system coordination. Step by step, they built a modular robot system – a process that required not only creativity and innovation but also patience and trust.
Embracing failure is integral to the learning process. Students only grasp a technical system’s true complexity when they push it to the limit, Zeilinger says. What matters, she argues, is understanding why failures occur and then making targeted improvements. These iterative refinements are what build an engineering mindset.
Hernández Oendra sees this as the format’s greatest strength: “Projects like Swarm don’t merely impart knowledge – they teach students to break down complex problems, shoulder responsibility and work effectively as a team.” The students all display remarkable motivation and eagerness to learn, he adds. And the challenges they encounter help them grow both academically and personally.
Zeilinger often sees former focus project students return to her lab. “We’re delighted when they keep in touch,” she says. “Some even choose to write their Bachelor’s or Master’s thesis within our group. It shows just how lasting the impact of these projects can be.”
Hernández Oendra agrees. “Many students gain more from these projects than they initially expect,” he says. “Afterwards, they approach new challenges with a different mindset. They’re more confident, solution-oriented and open to collaboration. And that’s exactly what this kind of learning is designed to foster.”
Focus projects
The Department of Mechanical and Process Engineering at ETH Zurich introduced the focus project format over two decades ago. This coming spring, 18 teams will present the prototypes developed during a year of intensive project work.