Learning from a robot

Utrecht students build an underwater drone

Underwater robots are meeting a rapidly growing need. At Utrecht University of Applied Sciences, they form the link between a range of technical degree programmes, which enable students to learn by doing.

TEXT: PANCRAS DIJK

They catch your eye the moment you walk in. Three underwater robots are lined up on a long table in the practical training room of the Smart Systems for Healthy Living research group at Utrecht University of Applied Sciences. It all began with the one at the front, which is still quite small. The other two are clearly more advanced and heavier in design. The Robosub team took part in an international competition this summer with the latest version. For the second year running, they returned with the award for the most innovative team. ‘We need to shave off five kilos next year,’ says team leader Rik Lafeber. ‘The students fought hard this year to ensure it didn’t have to be any smaller. But now they’re saying themselves: this needs to be lighter, that needs to be more compact. So it’s definitely going on a diet. That’s also great input for the new projects.’

The Robosub team has only been in existence since 2024, but is developing at breakneck speed. Whilst many student teams largely start from scratch every year, each intake of Robosub picks up where their predecessors left off. They are usually full-time team members for one semester, but often remain involved afterwards. The enormous workspace, says lecturer Matthijs Smakman, forms the ‘beating heart of the research group’ on the campus on the east side of Utrecht. That research group comprises three main disciplines: mechanical engineering, electrical engineering and ICT. ‘And in some projects, these disciplines intersect very clearly. Certainly in the underwater robot project, we absolutely need each of those three disciplines,’ says Smakman.

A safe living environment

Lafeber explains how it all began. ‘We’d been conducting research into additive manufacturing here for a long time. But that technology has now become commonplace and is therefore no longer particularly interesting to us from a research perspective.’ Together with team supervisor Colin Voncken, he looked for a new, exciting project that could keep students occupied for a longer period, yield relevant new knowledge and meet a real societal need.

Their attention turned to maritime engineering. ‘There’s so much going on in that field at the moment. From the accidental or otherwise intentional striking of cables on the seabed to the new European regulations on water quality and the inspection of quays, bridges and locks,’ says Lafeber. There is also a great need for automation in this field, whilst education has so far paid hardly any attention to it. ‘Companies such as Boskalis have to train their own staff in this field. That’s strange, because the number of companies in our country involved in water, water quality or shipping is quite large.’

Voncken describes the ultimate aim of the Robosub project as follows: ‘the design, construction and testing of underwater robots that make a significant contribution to a safe, clean living environment.’ That safety could, for example, relate to bridges that traffic can cross safely, explains Voncken, whilst ‘clean’ could refer to water quality or healthy biodiversity.

A modular approach

Version one of the Utrecht team’s underwater robot was primarily a proof of principle. Version two was capable of a little more and was fitted with an arm. Version three, which recently took part in the European Robotics League in Norway, is largely modular in design.

There is also a practical reason for this modular design, explains Voncken. The students all work in small teams, each on a specific component: a hydraulic system, the battery pack, communication electronics, a manipulator, software or image processing. By keeping these modules as independent as possible, the project remains manageable and safe. ‘We don’t want anyone to accidentally touch a cable, break something and then have no idea where the fault lies,’ says Voncken. ‘So we isolate functions. Everyone works on their own module, which later fits back into the whole.’

The result is a remarkable machine. The robot can be adapted to different tasks by replacing entire modules. Today, it has a gripper arm attached. Tomorrow, that could be a drill or a measuring system. The ‘brain’ remains the same and is ultimately intended to function with increasing autonomy.

Specific research questions

Although the Robosub project is all about building cool underwater robots, it is certainly underpinned by a real need from industry and government, explains lecturer Matthijs Smakman. Water quality is moving ever higher up the political agenda: autonomous robots can play a major role in monitoring it. Many quay walls, as well as bridges and locks, are approaching the end of their theoretical service life. Underwater robots would be ideally suited to carrying out regular, thorough inspections. ‘Essentially, you want to carry out a diving operation without constantly needing a team of divers,’ says Smakman. ‘Not only because of staff shortages, but also for safety reasons.’

Added to this were enquiries from water authorities, local councils and port authorities. How do you inspect riverbanks? How do you monitor water quality? How do you detect leaks? How do you monitor biodiversity? These are all applications where small, autonomous underwater drones can provide a solution.

Instead of a traditional research project, the research group therefore decided to fully integrate research and teaching. Students help build a robot whilst simultaneously working on practical research questions drawn from the field. This means that virtually every degree programme makes a contribution. Mechanical engineers design structures and hydraulic systems. Electrical engineers develop circuit boards, power electronics and communication systems. ICT students write software, build simulations and work on autonomous navigation. Industrial engineers focus on planning and organisation.

Small core group

The underwater robot serves as a common platform for this. ‘An underwater robot involves hundreds of questions like these,’ says Lafeber. ‘How thick should a wall be? How do you make something watertight? Which materials do you choose? How do you control a motor? Every small question can be linked to students’ learning objectives.’

This approach makes it possible to involve students in the project in a wide variety of ways. Some work for a few weeks on a specific assignment within a course. Others get involved through multidisciplinary projects. A small core group even remains involved for several years as junior researchers.

Lafeber picks up a thick folder. It turns out to be filled with wiring diagrams, soldering instructions, working drawings, revisions and test protocols. The wiring alone fills twelve folders. ‘We build it in such a way that we can build it again straight away. We document everything.’

The team makes virtually every component themselves. Each component is given a number, a working drawing and a fixed place in the documentation. According to the team, the current robot contains around 3,500 components that they have designed and produced themselves.

‘We’ve made everything ourselves,’ says Voncken. ‘Apart from the camera, the on-board computer and a few specific parts.’ That emphasis on documentation serves a second purpose. Students learn not only to design, but also to work in a way that’s transferable. That turns out to be at least as important as the building itself. Because at the end of every academic year, a large part of the team moves on, although most do remain involved. ‘Former students regularly pop in for an evening to help out,’ says Lafeber. ‘It’s really starting to become a community.’

Upward momentum

This was also noticeable in Norway, where the team took part in an international competition this summer. During rehearsals for the final presentation, the alumni were even stricter and more critical than the lecturers. ‘Colin and I looked at each other and thought: two years ago, you were doing exactly the same as the teams you’re now criticising,’ says Lafeber.

For the competition in Norway, thirteen student teams from various countries brought their underwater robots to carry out tasks derived directly from maritime practice: picking up objects, inspecting infrastructure, following a pipeline, locating an underwater charging station and docking at it. For Robosub, it was particularly challenging. The robot is designed for Dutch inland waterways, where visibility is almost always poor. Although the water in Norway is clear, the salt content meant that the buoyancy acting on the vessel was around 20 per cent greater than in the test tank at the university.

‘We’d done the calculations,’ says Lafeber. ‘But ultimately, you can only be sure it will work once the robot is actually in the water.’ To be on the safe side, steel discs were still carried along as ballast. Other teams could barely get their robots underwater. Some had to run their propellers continuously at full power just to stay submerged. It was precisely here that one of the Utrecht team’s own innovations proved its worth. The robot has an active buoyancy system: instead of adjusting buoyancy with cable ties and loose weights, the vessel can pump water in and out as it sees fit.

A real asset

That same capability also helps to maintain balance when various modules are suspended beneath it. During a challenge in which the robot had to remain on an underwater platform for ten seconds, that system proved its worth. ‘We had it pumped full until it sank rapidly to the seabed,’ says Lafeber. ‘Once it had landed on the platform, we could simply let go of the controls, whilst the other teams struggled to keep their robots in position underwater.’ The robot remained motionless on the platform, and the judges thought it had failed. ‘They shouted: “You’re giving up.” “Just wait,” we said. “We don’t need to hold it in place to keep it there.”’ Technical solutions like this are great for showing off in a competition. But they also have serious, practical uses. Companies, for example, want to be able to leave underwater robots in one place for extended periods to take measurements, or to let them recharge autonomously underwater without constantly having to return to the surface.

Meanwhile, things certainly didn’t go entirely to plan in Norway. For a start, the robot filled with water almost immediately, including all the electronics. Months of work seemed lost in an instant, but the team members showed their resilience. Some of the students stayed at the competition to observe the other teams. The rest moved to the basement of the Airbnb, where all the spare parts were stored. ‘We worked late into the evening, but with half the team we managed to get the underwater drone working again.’

Emotional support

Despite its modular design, the robot is now so complex that virtually every problem affects multiple disciplines. A leak is not just a mechanical problem; it can also affect the electronics, software and control systems. Repairs are therefore always carried out in multidisciplinary groups. ‘It’s never a single problem,’ says Voncken. ‘Everything is connected to something else.’

The role of the supervisors goes beyond that of mere technical experts. They arrange sponsorship, seek out assignments, organise travel, link projects to subjects and learning objectives, mentor students and keep an eye on the bigger picture. ‘And as well as being process leaders, we sometimes also act as emotional support.’

For example, when another competition in Italy was suddenly cancelled. An alternative had to be found within a day. ‘The most senior team members were called in. “What do you think would be a good plan?” Less than an hour later, comparisons between different competitions were already on the table. The team opted for Norway.’

One of the project’s guiding principles is: fail early. Voncken: ‘First, build something out of wood. Test the idea. Realise it doesn’t work. Have a laugh about it and carry on.’ It is precisely by making mistakes early on that costly failures are prevented later. In this way, failures do not become frustrating stumbling blocks, but rather part of the design process.

That attitude proves to be at least as valuable as technical knowledge. The tutors cite examples of students who, three years ago, barely dared to give a presentation due to a fear of public speaking or a fear of failure. Now they present their designs to international juries without hesitation.

Passion

The team’s alumni often find good employers quickly. They work for the Ministry of Defence, shipbuilders or elsewhere in the maritime sector. Some still come back to help new students.

Perhaps that is ultimately the project’s greatest benefit. The robot itself will undergo further changes next year, with lighter components, improved software and new modules. Some ideas are discarded, whilst others develop into scientific publications.

But at the same time, a new cohort of students is arriving. ‘In this project, they come into contact with things that, unfortunately, have been somewhat cut back in the education system,’ says Lafeber. It is precisely from these technically challenging, practice-oriented projects that the most can be learnt. ‘It’s about soft skills gained through working in a team, but it’s also in the field of technology that they find common ground,’ says lecturer Smakman. ‘Whether they’re IT specialists or mechanical engineers, they’re all becoming engineers, with their passion for technology as their common ground.’