No retirement for Lennart Ljung

Last summer, Lennart Ljung celebrated 35 years at Linköping University. Down the years, he has been one of the university’s best-known profiles in research, both at home and abroad. He’s 65, but retirement holds no interest.

On the other hand, what does hold Ljung’s interest is a completely new research project where he’s returning to where he began his career with system identification, building models of complex systems.

Ljung has received an Advanced Grant from the European Research Council, awarded to especially outstanding, excellent researchers within the European Union. Over the next five years, he will have the opportunity to delve deeper into how identification methods can be improved, by examining how both non-parametric models and models with many parameters can be used more systematically.

“This could provide greater advantages and models with higher quality,” he says.

Ever since Ljung came to Linköping University's Institute of Technology from Lund in the summer of 1976, automatic control engineering has retained a place as one of LiU’s regular spearhead efforts.

68 doctors have been examined over the last 35 years, among them former LiU Vice-Chancellor Mille Millnert.

“We started early in system identification research, creating models of systems, and quickly occupied a good position internationally.” However today, the cutting edge is more to do with sensor fusion. It’s an area where automatic control engineering, with Professor Fredrik Gustafsson, has a strong position, and where we now have the most doctoral students,” he says.

Sensors of different types, like cameras, temperature sensors and pressure gauges, and gyros, are found in a number of everyday products, in cars as well as cell phones, and they are becoming less and less expensive.

“Poorer precision in the mechanics of the inexpensive sensors must be compensated with software,” he argues.

To increase security in interpreting the results, the same measurement values and signals are taken from several different sensors. One of the many concrete applications being researched just now is care that would be able to perform an automatic evasive manoeuvre if, for example, the driver has fallen asleep. (Car radars spot the danger)

But there is also a research breakthrough Ljung would like to see, which has been the basis for several of the larger strategic efforts of the last few years: putting computer science and artificial intelligence together with automatic control engineering and sensor informatics. In addition, visualisation, which lies reasonably close to automatic control engineering.

“The goal is to combine various sciences into its own cross-disciplinary field, although we haven’t succeeded in going all the way yet. Today we work with the same data, but from different starting points. I thought that we’d gone farther, but no one else in the world has been completely successful either,” he says.

A symbiosis of the three sciences would mean significantly greater opportunities to analyse and control large, complex technical systems such as unmanned aircraft, autonomous cars, or advanced industrial processes.

“We’ve come a long way, something seen in things like successful joint Swedish Foundation for Strategic Research applications (SSF), but I think we can go even further.”

From the 1980s onwards, Ljung was the coordinator and driving force in a number of large strategic efforts, with millions in research funding. Efforts with acronyms like ISIS, MOVIII, CADICS, ELLIIT, and LINK-SIC (see at right); several of them with strong industries as participants and, in some cases, as co-sponsors.

Industrial relevance has, of course, always been the be-all and end-all.
“As an engineer, you want to solve problems and see the technology be of use. This is a collaboration that yields patents and better products for industry, but we’re not consultants,” he explains.

Few have inspired respect in Swedish industry like Ljung. He is extremely clear that collaboration is not about sponsored research. If you want to have a share in outstanding research results, you need to have a deep understanding of the research carried out by doctoral students, and that it therefore requires time, patience, and opportunities in scientific publications.

“We’re more than glad to listen to industry, and they’re always on the steering groups, but I’d prefer it if people in there have doctorates themselves so that they know which conditions apply,” he argues.

It’s a strategy that’s always been more than successful, for both academia and industry.

Doctoral students and undergraduates are, and have always been, important. Ljung was Inspector in LinTek, the Institute of Technology’s student union, for more than ten years, and over the last few years has conferred doctorates on the Institute's graduates. In 2007, the departments for automatic control engineering and vehicular systems won honours for outstanding education environments from the Swedish National Agency for Higher Education.

“We’ve got a good reputation among the students, and several teachers have received their “Golden Carrot” awards from LinTek. We’re not just here for the sake of research; taking the technology seriously is important,” he says.

When I ask him what he’s most proud of after all this time, the answer is precisely what’s important for the students: Textbook on System Identification, one of twelve books he’s authored, the first edition of which came out in 1987 and which is still used around the world. Especially “The System Identification Toolbox”, developed for Matlab, a programming language. It’s software that’s used at a large number of universities and enterprises across the globe. Ljung is constantly working to develop the toolbox, which is successfully being sold and marketed by the American company Mathworks.

Who knows whether the work will receive a further boost now that he’s back, under full steam, where he started; with system identification?

Related Article

Car radar spots the danger 

Related Links

Lennart Ljung 

EU Advanced Grants 

Institute of Technology at LiU 

Department of Automatic Control, Department of Electrical Engineering, LiU  

Mille Millnert 

Fredrik Gustafsson 

(SSF) The Swedish Foundation for Strategic Research applications

LINTEK: The Institute of Technology’s student union 

Swedish National Agency for Higher Education 

Matlab at Mathworks