IVA’s Great Gold Medal – Father of a dozen professors

Thursday 25 October 2018

lennart-ljung.jpg

Professor Lennart Ljung is receiving IVA’s Great Gold Medal for his fundamental achievements in the area of control theory, ranging from deep theoretical contributions to concrete industrial applications. He has developed new methods to build mathematical models for industrial (and other) systems and has also developed software that is used throughout the world. He has worked closely with industry, giving many Swedish companies a competitive advantage internationally.

Lennart Ljung says that as a researcher he is a “child” of Karl Johan Åström. But he also has ten such “children” of his own in the form of researchers who are now professors and who started their career with him. Lennart Ljung has been an important force in control technology research in Sweden.

Lennart’s significant imprint on the research community is obvious when you visit the Department of Electrical Engineering at Linköping University. One corridor leading to the department is called “Ljungeln” (Ljungle in English). On an office wall is a picture from the “Forever Ljung” symposium in 2006. All proof of the respect he has earned over the years, combined with student humour.

“The PhD students made the picture when I turned sixty, at the same time as we held an international symposium. But it of course comes from my name,” says Lennart Ljung when we meet at the university.

Lennart is still active in research and says that, if anything, he has more time for it now after retiring and taking on a senior role. His desk is covered with scientific papers (according to Google, he has been cited more than 60,000 times as a researcher). The walls are adorned with the various distinctions and honours he has received.

Another interesting fact in this context is that Lennart is the only researcher in control technology to have won the top award from the International Federation of Automatic Control (IFAC) twice – once for his theoretical work and once for practical applications. In other words, he is a researcher who has advanced in his field from the rudiments to concrete innovations. Mathematics or physics were not enough for Lennart with his wide range of interests.

“Maths is, for example, very one-dimensional. You have to be really good at it for there to be any point; to stand out in the ranks. Engineering is a broader field; there are so many different paths to take.”

Lennart wanted to be a physicist when he was young and his thesis project focused on physics. But later he realised that he found physics boring.

The optimal field for Lennart proved to be somewhere in between – or above – maths and physics. At the end of the 1960s cybernetics had become popular and Lennart was attracted by the subject. Cybernetics involved control technology and constant feedback from both machines and living creatures. There was a multitude of potential applications, such as human-machine interaction, artificial intelligence (AI), information theory and much more.

“Control technology itself is a mathematical subject I think. You build and use mathematical models to make good decisions.”

Today, autonomy is an entire application area and Lennart emphasises the importance of the Wallenberg AI, Autonomous Systems and Software Program (WASP) for development in the field. WASP is a major nationwide initiative that can be seen as a continuation of the centre of excellence (“Industrial Control and Information Systems”) that Lennart created with funding from NUTEK in the mid-1990s.

But although control technology may be thought of as broad and boundless, Lennart stresses the importance of researchers, developers and entrepreneurs having a strong foundation to stand on and to build on.

“I’m probably a theorist deep down. But I think that even people who are super-theoretical should work on concrete applications to achieve a good balance in their research. I’ve tried to have that as my objective.”

Lennart studied with the celebrated Karl Johan Åström (born 1934, engineer and now a senior professor in control technology in Lund). Control technology became a separate discipline at Lund University in 1965. Åström was the first professor in the subject there.

“Today control technology holds a very strong position in Sweden, and that is largely thanks to Karl Johan Åström. Almost all professors in control technology are his ‘children’ or ‘grandchildren’.”

Several of Åström’s “grandchildren” are essentially Lennart Ljung’s “children” as well – he has supervised around 80 PhD students in the field and at least ten of those have become professors.

Lennart is also an international authority in the field. He started studying control technology at an international level early on. For example, he did his post-doc under Thomas Kailath at Stanford University in the mid-1970s before returning to Sweden to become a professor in control technology in Linköping.

Stanford is often held up as a model for its close connections between research and entrepreneurship. Was that the case when you were there?

“Yes, back in 1974–75 when I was at Stanford they were broadcasting classes in Silicon Valley via radio link on the computer. Spin-off companies in Silicon Valley often have their roots in Stanford, and Stanford has preserved its teaching role there.”

A very early example was the now popular MOOCs (Massive Open Online Courses). Lennart has worked very closely with computers in his research, although he doesn’t think that control technology as a subject is dependent on computers.

“But you can do a lot with the subject if you apply it through computers. You can build an autopilot system for an aeroplane, but it’s no good without good hardware in the plane to make it work.”

Lennart started using his solid background in control technology theory in 1986 when he created the System Identification Toolbox for the Matlab computer program. The toolbox enables users to build a model of a process based on measured input-output data (so called “identification”) in industrial applications.

One important application area is simulation and modelling, including within control design, diagnosis and monitoring, signal processing and even weather forecasting. When accurate modelling is used, for example, before a manufacturing process is launched, significant savings can be made in industry.

“This is how research can in the end become a concrete solution. If we create software, people can use our theories in different contexts. So we’ve taken our research from deep theory to application, and people use it without thinking about where it came from.”

Lennart has more strings to his bow than just control technology. This Linköping professor also has a degree in Russian from the time he was studying civil engineering in Lund. He did his military service at the interpreter school and has kept up is language proficiency.

“After a couple of shots of vodka I can get quite talkative in Russian.”

Lennart Ljung also lived for six months in Moscow as part of his PhD programme and made good use of his language skills there. The Soviet Union had been prominent in several research disciplines for a long time, not least in applied mathematics, and it was common for Swedish researchers to travel to the east to improve in those subjects. When Lennart arrived there it was 1972, Brezhnev was in power and the Soviet Union was in a state of decay.

Rather surprisingly, Lennart sees some similarities between his research field and the Soviet Union as a country.

“They thought they could control the economy in the Soviet Union with control technology methods. The five year plans were a control technology idea. But they had no feedback like we have in control technology systems, and that’s why their system crashed.”

Not surprisingly the enormous national control technology experiment that was the Soviet Union collapsed a few decades later. Today researchers in control technology are looking even farther to the east. Lennart returned recently from a trip to China.

“They ‘ve not only caught up with us in their high tech applications, they’ve overtaken us. China, for example, has a highly developed high-speed rail system which they created from nothing in a short space of time. Here we are dithering about having high-speed rail with a timeframe up to 2040.”

But Lennart Ljung also points out the negative aspects of the Chinese approach, such as a one-party system and a rectilinear decision structure. In China there are no veto powers for municipalities or other actors impacted by a large-scale project.

Efficient control technology systems – whether they are used in individual technical installations or large-scale government development and planning – require feedback.

LENNART LJUNG

Age: 72

Education: Bachelor’s degree in mathematics and Russian, 1967, Lund University. MSc in Engineering Physics, 1970, and PhD in control technology, 1974, both at Faculty of Engineering (LTH), Lund University.

Career: After a post-doc at Stanford University Lennart Ljung became a professor in control technology at Linköping University in 1976. He has been a guest professor at Stanford, MIT, UC Berkeley in the USA, at the University of Newcastle in the UK and in Australia.

Distinctions: Honorary doctorate at Uppsala University, Helsinki University of Technology, Baltic State Technical University in St. Petersburg, Université de technologie de Troyes, France, and Katholieke Universiteit Leuven, Belgium. Member of IVA since 1985, the Royal Academy of Sciences (KVA) since 1995 and the National Academy of Engineering (NAE) in the US since 2004. Lennart Ljung has been awarded the Giorgio Quazza Medal and the Nathaniel B. Nichols Medal, both from the International Federation of Automatic Control (IFAC). He has also won the Hendrik W. Bode Lecture Prize and the IEEE Control Systems Award, both from the IEEE Control Systems Society.

Author: Joakim Rådström