The AI of digitalization

This article is the last of four where I explore different dimensions of digital transformation. Earlier, I discussed business modelsproduct upgrades and data exploitation. The fourth dimension is concerned with artificial intelligence. Similar to the other dimensions, our research showed that there’s a clear evolution path that companies go through as they transition from being traditional companies to becoming digital ones (see the figure).

In the first stage, the company is still focused on data analytics. All data is processed for the sole purpose of human consumption and interpretation. At this point, things are all about dashboard, visualization and stakeholder views.

Evolution of the use of AI technology.

In the second stage, the first machine learning (ML) or deep learning (DL) models are starting to be developed and deployed. The training of the models is based on static data sets that have been assembled at one point in time and that don’t evolve unless there’s an explicit decision taken. When that happens, a new data set is assembled and used for training.

In the third stage, DevOps and MLOps are merged in the sense that there’s a continuous retraining of models based on the most recent data. This data is no longer a data set, but rather a window over a data stream that’s used for training and continuous re-training. Depending on the domain and the rate of change in the underlying data, the MLOps loop is either aligned with the DevOps loop or is executed more or less frequently. For instance, when using ML/DL for house price prediction in a real-estate market, it’s important to frequently retrain the model based on the most recent sales data as house prices change continuously.

Especially in the software-intensive embedded systems industry, as ML/DL models are deployed in each product instance, the next step tends to be the adoption of federated approaches. Rather than conducting all training centrally, the company adopts federated learning approaches where all product instances are involved in training and model updates are shared between product instances. This allows for localization and customization as specific regions and users may want the system to behave differently. Depending on the approach to federated learning, it’s feasible to allow for this. For example, different drivers want their adaptive cruise control system to behave in different ways. Some want to have the system take a more careful approach whereas others would like to see a more aggressive way of breaking and accelerating. Each product instance can, over time, adjust itself in response to driver feedback.

Finally, we reach the automated experimentation stage where the system fully autonomously experiments with its own behavior with the intent of improving certain success metrics. Whereas in earlier stages, humans conduct A/B experiments or similar and the humans are the ones coming up with the A and B alternatives, here it’s the system itself that generates alternatives, deploys, measures the effect and decides on next steps. Although the examples in this category are few and far between, we’ve been involved in, among others, cases where we use a system of this type to explore configuration parameter settings (most systems have thousands) in order to optimize the system’s performance automatically.

Concluding, digital transformation is a complex, multi-dimensional challenge. One of the dimensions is the adoption of AI/ML/DL. Using AI is not a binary step, but rather a process that evolves over time and proceeds through predefined steps. Deploying AI allows for automation of tasks that couldn’t be automated earlier and for improving the outcomes of automated processes through smart, automated decisions. Once you have software, you can generate data. Once you have data, you can employ AI. Once you have AI, you can truly capitalize on the potential of digitalization.

In-depth optics training keeps both students and the teacher sharp

Modern optics for optical designers trainer Stefan Baumer
Deep at his core, Stefan Bäumer is an optics fanatic. He finds great passion in teaching because he likes to spread knowledge and, as he says, it keeps him on his toes. With undiminished enthusiasm, he has been providing the optics trainings Modern Optics – Physical Optics and Applications and Modern Optics – Optical Systems Design (CMOP-2) at High Tech Institute for years. The training is tough – covering all aspects of optics – but also very valuable, participants tell him afterward.

 One of the nice things about the optics, Stefan Bäumer thinks, is the visual. “If you build a set-up in the laboratory with a light or laser beam and lenses, you can use your business card to follow the beam and see what happens to it and how it forms an image. I really like the fact that you can see the effect visually right away,” he says enthusiastically. He likes working in the field because optics are the heart of the optical (measuring) system, determining both the functions and the tolerance of the system.

“In the future, we will be moving more and more towards end-to-end modeling,” says Stefan Bäumer, lecturer at High Tech Institute.

Bäumer’s experience in optics goes back a long way. Already, during his Master’s in physics, which he followed at Washington State University, there was a link. Also, during his PhD at the Technical University Berlin, he was involved in optics at the Optical Institute.

After completing his PhD, Bäumer started his career at Philips in Eindhoven. Here he started as an optical designer at CFT and later at Philips High Tech Plastics. After that, he worked for eight years as a senior optical system designer at Philips Applied Technologies and Philips Research. After a short time, as senior principal engineer at Philips Lighting, TNO asked him if he wanted to join their team. He was happy to do so; it allowed him to turn on his light at another organization. After a career of seventeen years at Philips, he switched to the optical group at TNO in Delft.

Now Bäumer has been working as a senior optical designer at TNO for almost eight years. Since 2015, he has also been part of the principal scientists’ group, who help determine technical policy. As a principal scientist, Bäumer is co-responsible for the direction of research in optics at TNO.

Progress

To be able to determine which way to go with TNO Optics, it is certainly useful that Bäumer has been in the field for a long time. As a result, he is well informed of developments. “Actually, there are a number of important developments that have led to strong growth in optics. The accuracy with which you can model optical systems has increased enormously. The integration with other disciplines has also greatly improved. In addition, the making of optical elements has improved, because the manufacturing technology has made huge leaps forward. Finally, near-infrared possibilities for detectors and light sources have been added, which are used, for example, in medical research,” explains Bäumer.

Far-reaching technological developments in the field of optical system design have indeed significantly improved the performance of all kinds of simulation programs in recent years. Optics also benefit from this. “I used to work with rather rudimentary optical design programs with which you could model systems and create a layout. There was still a lot of manual work,” says Bäumer. “Nowadays, you can model much faster and more accurately. You can include all kinds of phenomena in your simulations, such as nanostructures and diffraction. The integration with other disciplines such as thermal disciplines, mechanics and the improved communication between software systems of the various disciplines have also led to progress.”

What is certain is that the more sophisticated manufacturing technologies now available have also greatly accelerated optical developments. This is due in part to lathes with much higher precision – diamond turning allows you to create incredibly precise optical surfaces, including free-form surfaces – and new techniques such as magneto-rheological finishing (MRF) and ion beam figuring (IBF). These techniques allow opticians to design optics with much better specifications. This has also led to the emergence of free form optics in the last ten years. This branch of the field designs new optical elements that have no translation or rotation symmetry over the optical axis. This offers room for better performance, miniaturization and new optical functionalities.


Stefan Bäumer: “I would like to give my students a good basic optical knowledge, which enables them to make a good estimate of what is possible with optical systems. And, of course, also where the boundaries of feasibility lie'” Photo: TNO.

'I mean that we are going to predict system performance, from source to detector, under all circumstances using high-performance computing, among other things, as is done with ray tracing via graphics cards.'

 In the future, Bäumer predicts that we will be moving more and more towards end-to-end modeling. Bäumer: “By this, I mean that we are going to predict system performance, from source to detector, under all circumstances using high-performance computing, among other things, as is done with ray tracing (calculating how light rays behave in the optical system, AB) via graphics cards. This is a technique that they already use in the film industry, but it is also gaining ground in the scientific field. I also expect that more attention will be paid to computational optics. Because more and more computing power will become available, there will be more possibilities to find a better compromise between optical hardware and data processing via software. Because of this, different choices will be made in optical systems. Also, nanostructured surfaces and materials will increasingly find their application in optics. For optics in general, developments in the quantum field will unlock a whole new domain.”

Total overview

With all his knowledge about optical systems and the developments in the field, Bäumer can tell and show his students a lot during the training courses. The training courses Modern Optics – Physical Optics and Applications and Modern Optics – Optical Systems Design (CMOP-2) cover all the basic principles of optics. “It is important for trainees to understand how things work and which principles are behind them. How do electromagnetic waves and diffraction affect optical systems? What are the polarization effects? These questions all pass in review,” emphasizes Bäumer.

The fact that this training gives an overview of the entire field of optics makes it unique. However, there are many specialist optics courses, which zoom in on a specific area such as non-linear optics or optical design. What makes this training unique, however, is that it covers the entire domain of optics. A team of no less than seven qualified instructors, each working in the field of optics and specialized in a specific sub-area, provides high-quality training.

“Precisely the breadth and amount of homework that students have makes it a tough training. But that homework is the key to becoming familiar with the subject matter. When I talk to the students afterward and ask them what they thought of it, they say that it was tough, but that they learned a lot,” highlights Bäumer. “The training is also very strenuous for me. Every training takes a lot of preparation and there is a lot of after work, but I also learn from it every time and that keeps me on my toes. I want to give my students a good basic optical knowledge, which enables them to make a good estimate of what is possible with optical systems. And, of course, where the boundaries of feasibility lie.”

This article is written by Antoinette Brugman, freelance journalist and contributor to High-Tech Systems magazine.

Recommendation by former participants

By the end of the training participants are asked to fill out an evaluation form. To the question: 'Would you recommend this training to others?' they responded with a 8.1 out of 10.

“My PhDs weren’t allowed to leave without leaving something on the table”

Trainer of the Microelectromechanical systems (MEMS) training
A pioneer in the design of microelectromechanical systems (MEMS) with an additional passion for everything mechanical, a pragmatist and a very good teacher. That’s professor Bob Puers in a nutshell. He was chosen lecturer of the year in 2018 for his excellent MEMS training.

A curious course with overwhelming feedback from the trainees – that’s how Bob Puers describes the MEMS training course he taught in 2018 to a group of fifteen industrials from Pakistan. Puers: “It was held in China because of difficulties with the exchange of Pakistani. The trainees were all extremely willing to learn. I really appreciated this eagerness and also the particularly good interaction with the group. We had a lot of discussion on a very high level.”

In their feedback, the trainees said about Puers: “It was an excellent training both in terms of contents and presentation. The trainer was exceptional in answering questions raised” and “The professor’s way of teaching is extraordinarily good.” This positive feedback resulted in a review score of 9.8 out of 10 and the title “Lecturer of the year 2018.” Puers is modest about his contribution and points out that all the praise is probably due to accidental circumstances. However, when explaining his way of teaching and his knowledge about microelectromechanical systems (MEMS), it’s easily understood how he earned the title.


Bob Puers has always stimulated his PhDs to physically build a device. 

The scientist

Puers’ MEMS experience goes back to his study in electrical engineering. He was very interested in research and had a special passion for everything mechanical. When he came in contact with Raoul Vereecken, a urologist at the University Hospital in Leuven, he got involved in the development of portable, implantable medical electronics. He continued his career in this domain and started his own research group at the KU Leuven in 1988. Soon he had the disposal of his own cleanroom to fabricate devices such as pressure sensors, accelerometers and flow sensors.

'My PhDs weren’t allowed to leave without leaving something on the table.'

In his research, Puers focused on the application of medical implantable electronics and the development of technology to produce sensors – he’s always been motivated to develop devices and is working in a pragmatic way to realize this. If Puers knows a certain principle works, he doesn’t delve too much into the details of the theory but uses this knowledge to put it into practice and make new devices. And being a man of practice: he’s always stimulated his PhDs to physically build a device. As Puers puts it: “My PhDs weren’t allowed to leave without leaving something on the table.”

Puers continues: “In our cleanroom, I developed lithography and application techniques with our group of researchers. We made more sophisticated mechanical structures – on a miniature scale. The whole process of developing a very small mechanical structure, integrating it in an electronic component – to convert the mechanical signal into an electronic one – and finally building a sensor out of it – that still fascinates me.”

There have been many developments in Puers’ discipline. “Back in 1985, our group was one of the first to develop accelerometers. These devices were ground breaking at the time. Nowadays, accelerometers are integrated into commercial products like smartphones and cars at incredibly low cost. There are quite a lot of devices we laid the basis for, ideas that were taken over by the industry later on. So, we had to search for new research domains several times.”

MEMS developments are still ongoing. The current trends are far-reaching miniaturization and very low power consumption. This makes sense, for many sensors are applied in portable medical applications and thus have to be as energy efficient as possible.

The teacher


As a KU Leuven employee, teaching was part of Puers’ tasks.

He started as a teacher of courses in biomedical electronic systems. Later on, he also taught about MEMS production technology. These courses still form the basis of his MEMS systems training at High Tech Institute.

“It’s challenging to educate people and to get them excited about the science domains that you find fascinating yourself,” Puers explains. “You’ll never get them all interested. Only about one third to half of the university students get excited about the subject, the others only do what they’re told. However, High Tech Institute trainees are always people with specific interests who share my enthusiasm. They usually have some experience already, so we have a lot of detailed and specific discussions during the courses. I really like that interaction.

'I explain the possibilities and the impossibilities of MEMS, zooming in at system level.'

Puers started his MEMS training course for High Tech Institute in 2009 – being a specialist, he was asked to educate people about MEMS. His goal is to introduce his trainees to the domain. “I want them to know more about all the techniques that have been developed over the years to produce micromechanical systems. I explain the possibilities and the impossibilities of MEMS, zooming in at system level. About half of the MEMS training I spend on the instruments we have at our disposal to build a sensor or actuator. These are all necessary techniques, like etching, bonding, packaging and coating. In the second part of the training, I teach the trainees about all kinds of successful applications, like flow and pressure sensors, optical systems and medical implants. In the end, the trainees should know what’s possible and what’s almost impossible. I want them to be able to judge how realistic new concepts are.”

His vast MEMS experience, being involved from the very beginning, makes Puers a knowledgable teacher. But he’s also skillful in tuning to his audience. “I always answer questions that pop up during the course. Sometimes I can do that straight away because I know the answer from experience. If I don’t know the answer, I get back to the issue the next course day. I like to anticipate questions and feedback in my training. Teaching is a process of evolution. In every new course, I use the experience of previous courses, so my intellectual baggage as a teacher is continuously being enriched. In this way, I’m constantly refining my courses and adjusting them to my audience and their prior knowledge.”

This article is written by Antoinette Brugman, tech editor of Bits&Chips.

Recommendation by former participants

By the end of the training participants are asked to fill out an evaluation form. To the question: 'Would you recommend this training to others?' they responded with a 8.5 out of 10.