Linking optics to systems thinking and AI offers great opportunities for the Netherlands

Reading time: 6 minutes


Its multidisciplinary character and its high-tech aspects make optomechatronics a subject that fits TUE’s High Tech Systems Center like a glove. Adaptive optics, integrated photonics, artificial intelligence – Bits&Chips sat down with HTSC fellow Ton Peijnenburg to discuss the trends and latest developments.

“The combination of optics and mechatronics is extremely interesting,” says Ton Peijnenburg of the High Tech Systems Center (HTSC). “Mechatronics is a well-known strength of the region. Just look at all the high-end devices being developed here. Optics often plays an important role – to measure positions, to expose, to magnify objects. Optics is always a topic.”

The Netherlands doesn’t have a large optical industry like Germany or Japan. “We don’t produce optical components on a large scale,” admits Peijnenburg, “but we do know what we’re talking about and we definitely have something to contribute.” He notices that companies from these large optics countries, but also the US and elsewhere, find their way to the Netherlands when they have an optomechatronic challenge. “That’s largely because we’re good at systems thinking. After all, optics is often part of a larger system. Optical elements have a finite accuracy. If that’s not good enough, you’ll have to get it from somewhere else in your system, for example from very controlled movements.”

International companies find their way to the Netherlands when they have an optomechatronic challenge. Credit: Bart van Overbeeke

One of the routes the High Tech Systems Center is taking to stimulate the optomechatronics knowledge in the region, is through the Nano Opto-Mechatronics Instruments initiative (see box). It hasn’t yet come to an industry-wide consortium, which is the HTSC’s usual approach. Peijnenburg, however, sees plenty of leads. “Look at the ELT, the Extremely Large Telescope,” he gives an example. “To compensate for air vibrations, the mirrors have to be adaptive. Even with fairly large surfaces – the ELT’s M2 mirror has a diameter of more than four meters – you see that we can now do that actively as well. A ‘bed of nails’ is placed at the back of the mirror. Adjusting the height of the nails will slightly deform the mirror surface. We’re talking minuscule movements of less than a micrometer, so high-end optics and high-end mechatronics come very close together here.”

Peijnenburg also sees applications for adaptive optics outside astronomy. “Think of ASML’s lighting systems,” he notes. “And the end isn’t in sight yet because with more research and state-of-the-art actuation techniques, we can still gain a lot in terms of stability, speed, accuracy and adjustability.”

There’s a great opportunity for the region to be found on the interface between artificial intelligence and high tech. Credit: Bart van Overbeeke

Entrepreneur wanted

Adjacent to deformable optics are the hot developments in laser-based communication. “To give the beam between the ground station and the satellite the right characteristics, you can also work with adaptive optical elements,” argues Peijnenburg. “TNO is way ahead of the troops in that area. But I certainly see opportunities for the HTSC to contribute. For example, you need relatively traditional control technology to properly align those mirrors. That’s very close to the servo technology used to keep wafers neatly in place, which also involves high dynamics.” Peijnenburg expects that within two to three years, “very nice innovations” can emerge from these developments. “Also within high-end equipment.”

Peijnenburg mentions integrated photonics as a third focus area. Over the past six months, together with Ben van der Zon of High Tech NL, he investigated what companies in that sector need and what the missing functionalities are. “One of the challenges that emerged is probing,” he states. “At various points in the production process, you want to perform optical qualifications, for example to check whether the transmission in a channel is good. It’s far from trivial to develop probes for this. Those fibers not only have to adequately take in the light – a challenge in itself – but it also has to be done on a scale much smaller than the wavelength of the light.”

There’s a plan on the table to start a research project for this and Peijnenburg hopes to receive support from the corona recovery fund React. “That contribution is obviously very welcome, but in the end, it’s about putting something on the market,” he says. The HTSC consults with parties such as Photondelta, CITC and the new PITC on how to shape such an OEM, which company can take on that role and who the enthusiastic entrepreneur is who dares to take the step. “A process that takes time.”

It’s almost embarrassing how easily a robot footballer with artificial intelligence can recognize a ball. Credit: Bart van Overbeeke

Artificial intelligence

For the last six months, the HTSC has been officially part of EAISI, the Eindhoven research institute for artificial intelligence. Not exactly a place where you’d expect to find optomechatronic experts. Or is it? “There’s certainly a data processing component in the optical qualification of photonics chips,” responds Peijnenburg. “And the control of thousands of actuators in adaptive optics can be optimized with artificial intelligence. With smart AI techniques, you can also use software to make camera images even sharper.”

Peijnenburg pulls out another example. “It’s super easy to let the robot footballers of TU Eindhoven recognize objects with AI. It’s almost embarrassing when you compare it to traditional vision techniques. Anyone with a webcam, a decent processor and some AI knowledge can find the ball on the field better than we can with an industrial camera. If the color of the ball changes, that camera won’t even see it anymore.”

On the other hand, measuring exactly where the ball is, is a lot harder with AI. “The traditional techniques are better suited for that. In that respect, everything’s complementary. You have to think carefully about what to use and when,” advises Peijnenburg. “Without a doubt, there’s overlap between EAISI and the HTSC, especially if you apply artificial intelligence to high-tech systems. Those worlds don’t yet know each other very well, so that’s a great opportunity for the region.”

This article was written in close collaboration with TUE’s High Tech Systems Center.