Oral-History:Brian Anderson

About Brian Anderson

Brian Anderson, an IEEE Life Fellow, was born in Sydney, Australia. He completed his undergraduate degree in Mathematics (1962) and Electrical Engineering (1964) at Sydney University, and his doctoral degree in Electrical Engineering (1966) at Stanford University. After finishing his degree, he stayed at Stanford as an Assistant Professor, while simultaneously working in industry at Vidar Corporation. Following these posts, he returned to Australia as Professor of Electrical Engineering at the University of Newcastle (1967-1981). He then went on to the Australian National University in Canberra where he served as Professor and Head of the Department of Systems Engineering, and later as Director of the Research School of Information Sciences and Engineering from 1994 to 2002. Anderson also served as inaugural CEO of National ICT Australia in 2003 and Chief Scientist of NICTA from May 2003 to June 2006, as well as several visiting appointments around the world. He currently remains a Distinguished Professor of the Research School of Information Sciences and Engineering at the Australian National University.

Anderson's research interests include adaptive and control systems, automation and control engineering, circuits, signal processing, and communications.

In this interview, Brian Anderson discusses his memories, reflections, and thoughts in regards to his fifty year career in Control Systems. He provides an overview of his time at Stanford, Newcastle, and the Australian National University, and he outlines his own experiences in, and contributions to, the field.

About the Interview

BRIAN ANDERSON: An Interview Conducted by Rob Evans, IEEE History Center, 4 May 2015.

Interview #759 for the IEEE History Center, The Institute of Electrical and Electronics Engineers Inc.

Copyright Statement

This manuscript is being made available for research purposes only. All literary rights in the manuscript, including the right to publish, are reserved to the IEEE History Center. No part of the manuscript may be quoted for publication without the written permission of the Director of IEEE History Center.

Request for permission to quote for publication should be addressed to Oral History Program, IEEE History Center, 445 Hoes Lane, Piscataway, NJ 08854 USA or ieee-history@ieee.org. It should include identification of the specific passages to be quoted, anticipated use of the passages, and identification of the user.

It is recommended that this oral history be cited as follows:

Brian Anderson, an oral history conducted in 2015 by Rob Evans, IEEE History Center, Piscataway, NJ, USA.

Interview

INTERVIEWEE: Brian Anderson
INTERVIEWER: Rob Evans
DATE: 4 May 2015
PLACE: University of Melbourne, Australia

Early Life and Education

Rob:

My name is Rob Evans, today is the 04 May 2015 and I am delighted to be talking to Brian Anderson about his memories, reflections and thoughts on a very distinguished 50 year career in Control Systems. So Brian, thank you for talking to us today.

Brian:

My pleasure Rob.

Rob:

I guess the logical place to start is tell us what you are doing now and also your memories about your early days, university, going to Stanford , those early days.

Brian:

Well now I am a Professor at The Australia National University and a researcher in National ICT Australia. I’m part-time and I’m pretty close to retiring.

Rob:

[laughs]

Brian:

But I started my life in Sydney and I did most of my growing up in Sydney, a little bit in London and Brisbane and then ended up at Sydney University. And in Sydney University I studied Mathematics and Electrical Engineering. To get the equivalent of a Dean’s award in Electrical Engineering you actually had to study a lot more science, mathematics and physics, than for a normal degree and that was the route that I went.

Rob:

Ok. You clearly are an outstanding mathematician. Is the engineering part of the Dean’s award or you had also a passion for engineering, or?

Brian:

Well, I think from high school days I was very interested in maths and physics and I used to play with an electric train set and electric motors and repaired TVs. So I acquired some engineering interest and I had two close school friends whose fathers were electrical engineers and I admired them both very much. One actually was chairman of Electrical Engineering at Sydney Uni and I think my hobbies and my friendship with these boys and knowledge of their parents whom I admired, shifted me towards Electrical Engineering away from maths. And in thinking about it a little bit more I do think that mathematicians as a class are a bit more inward looking than perhaps my personality is, and I was more comfortable with the more outward looking personality of engineers.

Stanford University

Rob:

So you ended up going to Stanford. Brian, can you tell us about that?

Brian:

Yes, I did want to do graduate work and I sought advice from the father of the school friend who by now was Dean of Engineering at University of British Columbia in Vancouver, and he said definitely go to the United States and pick one of Stanford, Berkeley, MIT or Caltech.

Rob:

Fairly safe options.

Brian:

Yes, so I wrote to them and some professors at Sydney University agreed to be referees. And then one of these Professors said to me one day “I think I can get you a research assistantship at Stanford” and what was going on was there was a Stanford faculty member, Robert Newcomb, who was on a sabbatical in Australia and he’d met the Sydney University Professor and received the sales pitch about myself and agreed to at least meet me. I don’t think he’d agreed at that stage to the research assistantship. Anyway, I met Bob Newcomb and he actually gave me a small problem to work on, which I solved, and then he said “really I believe I can fix for you a research assistantship at Stanford.”

Rob:

Wonderful.

Brian:

So it was a very fortuitous and wonderful thing, I was so fortunate with that.

Rob:

And the environment at Stanford in those days, this is around about?

Brian:

‘64, 1964.

Rob:

1964. Can you tell us about that environment, who else was there?

Brian:

Okay, when I was in undergraduate school I used a number of textbooks and they were pretty well all American in origin and I arrived at Stanford and I saw the authors of half those textbooks. So, I was in some kind of nirvana, and it is true that the excellence at Stanford is so pervasive and general to the institution rather than confined to exceptional pockets, which was I think the case at Sydney University, that to move there from Sydney was an immense privilege.

Rob:

Yes, yes and so who were some of the people around with you?

Brian:

Well my own supervisor, Bob Newcomb who is still an active professor at the University of Maryland and then in the systems area Gene Franklin, a very famous person, Bernie Widrow, another one; Arthur Bryson, another one; and certainly Rudi Kalman. Rudi arrived as a new professor at Stanford the same quarter that I started my graduate study and I had the experience of enrolling for courses taught by Rudi in two successive terms.

Rob:

Oh that’s fantastic!

Brian:

It was.

Rob:

And you, with Bob Newcomb, and I remember studying his books at the University too on circuit theory. So you worked on circuit theory but you had a very Systems and Control Systems view of all that.

Brian:

Well, yes I did work on circuit theory with Bob. It was actually a theory of time-varying passive circuits. So circuits where you had inductors and capacitors and resistors and transformers but the element values could vary with time, and that was what my PhD thesis was about. My control systems interest at the time I left Sydney University was not that strong, because I don’t think the instructor that I had at Sydney University had been able to teach it in an exciting way and or the textbook was perhaps not as exciting; and what absolutely tipped the balance was attending the lectures by Rudi Kalman.

Rob:

Yes, I can imagine.

Brian:

It was a challenge to keep up with them and I enjoyed that challenge and was able to handle it and of course, as we all know, he was at the absolute frontiers of the discipline. So it’s very exciting to sit in on something like that and sometimes he’d be talking about recent research; and I’d say my move out of circuit theory into more generally control systems and systems theory is as much due to Rudi Kalman as it was to Bob Newcomb.

Rob:

And a lot of your work involves spectral factorisation and things like that in your PhD and I guess the move from Wiener filtering and all the spectral factorisation issues. So you were right in the middle of that move to differential versions of everything, which must have been very exciting.

Brian:

That’s right. There is a way of looking at classical passive network synthesis and of Wiener and even Kalman filtering and of linear optimal control using this tool called spectral factorisation.

Rob:

Yes.

Brian:

and I had different insights into spectral factorisation from my own reading plus working with Bob on my thesis and with Rudi.

Rob:

Yes, I remember reading some of your stuff on that. I was struck by the fact that from that perspective you could see all those advances in one shot and it was quite a remarkable thing.

Brian:

There is one more thing that I might mention. I had a discussion with Rudi one day about an unfinished aspect of some work he had done and he encouraged me to work on it. And I worked on it and wrote up something as a paper and he went through it and he tore that paper to shreds, not from the point of view of the technical content, but the writing style. And I learnt an enormous amount about technical writing, valuable stuff from being put through the wringer by Rudi Kalman.

Rob:

And I suppose many others were also put through the wringer by Rudi.

Brian:

They were; he is a tough guy.

Rob:

And also in that you mentioned people like Bernie Widrow, so that exposure to that exciting beginnings of adaptive systems was all there for you too, and...

Brian:

That’s right; he was a pioneer in the adaptive systems area well before really Narendra, Morse and so on, from the signal processing point of view.

Return to Australia

Rob:

Yes, but your contributions have spanned signal processing, control, communications, so it sort of all make sense when you hear that background, and what did you do after you finished your PhD?

Brian:

Well I stayed on at Stanford as an Assistant Professor and simultaneously I continued with a part time job I had in a company I had started [working at] while I was still a student, called Vidar Corporation that later went into Ling Temco Vought, I think and TRW. It was originally a startup done by a Stanford Professor, Malcolm McWhorter. And then after a year I returned to Australia.

Rob:

Okay, and why did you return? Obviously from that jumping off point you could have had anything.

Brian:

Yes, yes. Well, there is no question that the technical environment...

Rob:

We are glad you did by the way.

Brian:

[laughs]

The technical environment in America and in particular in Stanford could not be bettered, but even in those days I thought all other aspects of the life environment in which I was could be bettered by returning to Australia. I think it is a wonderful country and of course I had friends and relatives here. I have always considered myself a sort of passionate Australian and so it was a wish to come back. I did want to obtain a position where I could have the sort of freedoms that I had gotten used to at Stanford. I think there was a pattern in Australian Universities at that time to over-manage junior faculty and I would have found that very, very constraining so it wasn’t a case of taking any job in Australia, it was a case of wanting to come to Australia, but it had to be the right sort of job.

Rob:

And it was; you came back as a full Professor.

Brian:

And it was. Yes I came back as a department head, of a small department and in that role there were immense freedoms and there was the freedom to make some appointments remembering one thing I learnt at Stanford that was sourced, If you like, with Fred Terman, a very famous Electrical engineer.

Rob:

Was he at Stanford?

Brian:

Yes, he was. He had been Provost of Stanford I think at the time I was there and he had helped Hewlett & Packard get started. And I read some of the stuff he wrote and he said what you must have is steeples of excellence in adjoining or reinforcing areas, rather than have one person in control and one person in optical electronics or something. If you are going to do control you’ve got to have several people in control or if you’re going to do optical electronics you’ve got to have several people. So I’d acquired that lesson and I used it at Newcastle when I was involved in making further appointments there.

Rob:

And so that explains in a way how you built the spectacular group in Newcastle, Fred Terman is responsible.

Brian:

He gave the intellectual framework.

University of Newcastle

Rob:

Yes, yes I appreciate that. So Brian, you built a spectacular group at Newcastle and made some very strong appointments and it was a very productive and technically exciting period of your life. Would you like to tell us a little bit about that and reflect on.

Brian:

Well one of those was John Moore with whom I had a technical association and friendship over some decades, and he’d joined the University of Newcastle about a week or two weeks after me. We’d more or less concluded the arrangement walking around the gardens of the Allerton conference centre of the University of Illinois.

Rob:

So you were both still in the US at that time.

Brian:

Yes, he was at the University of Santa Clara. And let’s also be clear that you were also in fact at Newcastle, Rob and with Graham Goodwing, another one who’s a very famous person. Anyway John and I clicked very, very well technically and I think we were also motivated by the idea of book writing and that was very unusual here in Australia. So we did a book on Linear Optimal Control which of course drew very heavily on Rudi Kalman’s work, but we’d done some work in that area ourselves. And then we put a lot of effort into studying aspects of Kalman filtering and in particular smoothing and we’d become aware that there were really no smoothing algorithms around for which stability had been proved and in fact most of them had internal instability through unstable pole-zero cancellations.

Rob:

Yes.

Brian:

So with the graduate students and with each other we did a lot of work on smoothing and that plus our knowledge of Kalman filtering, besides finding its way in the papers, was included in the book we wrote on optimal filtering which is, I think, much more widely cited than the book on linear optimal control, and it is cited by people like econometricians, not just people in electrical engineering.

Rob:

Yes, okay.

Brian:

There is one other result that I really remember with John Moore, with a lot of affection almost, and it concerns decentralised systems, which of course are a big issue now. But you can imagine a system with two scalar inputs and two scalar outputs, a linear time-invariant system, and you can imagine that you might be told to stabilise it, because it’s initially unstable, with a controller that connects output one to input one and output two to input two. And there are examples of such systems for which it’s impossible to find time-invariant controllers with this decentralised structure which will stabilise them. What we were able to show is that for many such systems you could have a periodically time-variant controller and it would keep it stable, and I believe that there is an analogy with many social systems including systems in democratic countries which remain stable because governments change. The control.

Rob:

That is a very interesting connection.

Brian:

It is kind of far-fetched, but nevertheless I think there is value in this because any one government has strengths and weaknesses and if you leave them in place too long the weaknesses can really eat away at the institutions of the country and so on, and by changing governments or changing CEOs you get someone else or another team which has a different set of weaknesses.

Rob:

Yes, this is very interesting insight, I do remember that paper well of course, it was a lovely paper. Through that time at Newcastle you also kept up your circuit theory and wrote a rather beautiful book.

Brian:

Yes, well that’s nice of you to say that. It was, so to speak, my first love. I don’t know if you remember your first girlfriend, circuit theory was my first love and I wrote the book with an outstanding student who was originally from Thailand. I might say we have kept up our connection and he came to my 70th birthday party in Australia which was just lovely, and the King of Thailand was presented with a copy of that book. And it is widely cited today and I am proud of it and remember the work with him with great affection.

Rob:

Yes, I often look at that book too. I find it a remarkable book actually and do you have any sadness that circuit theory and synthesis and all those areas are sort of very rarely taught in electrical engineering courses these days. It was a lovely development.

Brian:

Yes, but I don’t have sadness. Life has to go on and part of life is renewal and there are other exciting things that have come in their place. I’d have to say I am critical that in my undergraduate electrical engineering course there was so much attention given to vacuum tubes and rather less if you like given to transistors and life hadn’t gone on enough for the undergraduate courses at Sydney University.

Rob:

Another result that I thought was remarkable from your time in Newcastle was your output feedback work. Do you have anything to say?

Brian:

[laughs]

Rob:

It was a tough area.

Brian:

Well we should recall what it is, right? You’ve got a linear system that is unstable and in general it’s got multiple inputs and multiple outputs, and one is looking for a constant feedback law not dynamic but static that when put round the system will stabilise it. And there is no simple solution to that problem but, what with Eli Jury and Nirmal Bose, we discovered was that it was a special case treatable by a general theory logicians had developed, particularly a fellow called Tarski, but there was a mathematician called Seidenberg that if you like provided a massive massive generalisation to Routh-Hurwutz criteria and you could answer the question, yes there is or no there is not, a stabilising feedback law and then with ability to solve multivariate polynomial equations if there was a feedback law you could find one, because it was given as a real solution of these multivariate polynomial equations. The answer was depressingly technical. It wasn’t a beautiful answer, it was a very technical answer. Nice to get the problem out the way, it’s true that many many people looked at the problem and solved special cases, but I think that’s the best you will ever be able to do with that problem.

Rob:

Well I recall a seminar in which you expressed exactly those thoughts about 40 years ago, is it?

Brian:

Yes, it is amazing.

Rob:

But where you said the problem was eating away at you, it was very very hard and you were looking forward to being able to solve it and you solved it but your solution you found disappointing in a way. I remember you saying exactly the same words so long ago, so you are consistent.

[laughs]

ANU

Rob:

So Brian, you moved from Newcastle around 1982 and went to ANU which was then in a sense starting up a new department and another enormous flowering of your technical work in adaptive control and many other areas. So can you tell us a bit about the move to ANU. ANU wasn’t so focused on engineering so that must have been a challenge in itself.

Brian:

So, I was the first engineering professor at ANU. And they stuck me into a physics school and said you are going to start a department of Systems Engineering. And the Physics School had some very traditional physicists within it, which is code for saying there were many physicists who didn’t really imagine engineering had much to offer intellectually and was in intellectual terms perhaps a pigmy next to the giant physics.

[laughs]

Brian:

So there was from some parts of the physics school clear hostility to us, because why couldn’t the money have been better spent on more physics? Any way you can live with that because we were a self-contained identity and as you were saying, I think, at that stage I started putting a great deal of effort into adaptive control. And while I was still in Newcastle working with Rick Johnson, who is now at Cornell doing other things, we’d established a sort of complicated condition to guarantee exponentially fast convergence and therefore robust behaviour in many cases of a number of adaptive control algorithms. Now when I moved to Canberra, shortly after I moved there, I heard a seminar by a French person visiting in which he described a laboratory set up using adaptive control in France which had exhibited massive misbehaviour after about a week of operations. The thing had gone on beautifully and then bingo.

Rob:

Probably another analog to governments there as well...

Brian:

And then I heard about such a phenomenon in some other place, I can’t remember where, and then I, using an apple II+ computer, in case you don’t know what its memory was, It was 64K, using an apple II+ computer in a language called ‘basic’ I had an idea why this was occurring and I simulated it at home. And I got this behaviour after 45 minutes, and that behaviour was explicable using the tools that I develop with Rick Johnson at Newcastle. And essentially if you don’t give enough excitation to an adaptive system and the adaptive system has got an implicit or explicit identifier in there in an insufficiently excited state, it will make a mistake with the identification or the identified parameters will drift round and you can drift to a region where you get instability and so violent oscillations. So that was also the notion of persistency of excitation and the mechanism whereby unexpected oscillations could occur, could really also explain what became known as Rohrs-Athans (I think) counter example. They asserted that no adaptive control algorithm could work, and that was an incorrect assertion of course. What they were really observing in their experimental work was what happened when there was an absence of persistency of excitation. There needed to be some un-modelled dynamics and so on along with it too in their example but it is essentially absence of persistency of excitation.

Rob:

And you’re a keen windsurfer and been keen for a long time, so I guess that’s where the name of windsurfer approach came from?

Brian:

The windsurfer approach to adaptive control certainly came from my thinking about how I learnt windsurfing. Now I know I am talking to one of Australia’s ace windsurfers though you are a bit slower these days, but there is a process of learning where you can do certain things though you can’t do the next stage, and you try doing the next stage and you fall over or you do it with great difficulty and gradually you master the next stage and then you go onto the next stage beyond that and learn or fall or with great difficulty master it and then and so on. So you progressively, partly through experiment at the boundaries of your abilities, push out what your abilities are. It is the same really as learning tennis or something like that and the windsurfer approach to adaptive control embodied a different approach to adaptive control than learning the numerator and denominator coefficients transfer function; it involved learning what was going on in the pass-band and gradually expanding the pass-band and designing the controller accordingly.

Rob:

It is a very appealing approach intuitively, it makes a lot of sense.

Brian:

And I really came at it through thinking about windsurfing. Windsurfing is a wonderful example of adaptive control for the humans who control it.

Rob:

And I wonder if there are lessons from you hinted at it before for you know the way governments behave or stabilisation of communities and societies. I wonder if there are also lessons from control back into and from you know both ways in this type of thing. Sometime around this time you also did quite a lot of industry work. You did work with Boeing and model order reduction work and things like this.

Brian:

So just to mention what the Boeing work was, they were taking 200 person years to design the pitch control system for their commercial aircraft and it took that time because essentially they had a trial and error method. And you might say well why couldn’t they use linear quadratic control, and the problem was if you used linear quadratic control you got a controller that was about 60th order because the plant was that sort of order and no manager would agree to implementing a 60th order controller, even if they had taken a course from Rudy Kalman. So Boeing were insistent on having a low order controller which you could sort of get a feel for and so that yes, that was the problem I work on and I had a Boeing engineer with me for a year in Australia and it later gave rise to a bunch of papers and I wrote a textbook with a Japanese professor on modelling and model and controller reduction. He’d been a visitor with me for a year in Australia too and that was his speciality in Japan.

Rob:

Yes I think this is a really important area because as you say even in the days of lots of computation capability and everything else in the end if you end up with a high order controller somewhere there are people involved and they don’t understand it or it is much harder to implement than you think or something.

Brian:

Well that’s right. At any time we implement a controller we cross our fingers that the un-modelled aspects are not going to bite us in the backside and if it’s a 55th order controller who knows. [laughs] A big possibility for getting things wrong in terms of truncation of coefficients and the like.

IFAC

Rob:

One of the big things that you got involved with in your time at the ANU was becoming president of IFAC, the International Federation Automatic Control. I think at the time you got involved maybe between 50 and 60 countries were represented on IFAC.

Brian:

Yes, probably about 50 nearer to 50 than 60, I can’t remember the exact number.

Rob:

But that must have been quite a large job and a leadership position of some substance.

Brian:

Yes, I served 12 years as an officer which was the standard career trajectory for someone who was going to be a future president. The system brought you in knowing that you are going to be a future president and you did that between your 6th and 9th year. I was also brought in with the injunction from the then president that lots of things needed to be improved and that they were willing to give me an unusual degree of latitude. And my first task as an officer was to chair the technical board of IFAC and the technical board, prior to my involvement in it, when it had a meeting would get bogged down in the discussions of meeting dates and may be fees. It didn’t really do anything technical. So with some very able vice-presidents of the technical board whom I got on side, including Lennart Ljung, who was my successor as the chairman of the technical board three years later, we revamped what the technical board did. We put some structure around the sequences of meeting, we got processes to evaluate the quality of a (meeting) proposal, we stopped focussing on dates, we got some feedback after meetings, and then we started focussing on the publications, having done something about the meetings, we started focussing on the publications.

The time I became vice-president of IFAC and chair of the technical board there was just Automatica and now you would know there is a range of IFAC affiliated journals. And sometime during this period we also establish agreements with the ACC, the European Control Conference, the Latin American Control Conference and the Asian Control Conference to do things in conjunction with them so to speak, which was a very good move and it helped stop any evolution of alternatives to IFAC. One of the challenges in those times and the dates are quite important, I became a vice president in 1984, the Soviet Union still existed, and the upper level positions of IFAC were so to speak divided by arrangement between east bloc countries and everybody else. And that meant that sometimes you didn’t always get particularly able people in the roles, people will come to a council meeting and have an interpreter. You really can’t have an easy council meeting if people have got interpreters whispering in their ears. It stops a certain spontaneity and creativity of the deliberations and people are unsure of themselves and so on. So that was an issue that we had to deal with. Mind you there were some excellent people from those east bloc countries, Boris Tamm and Tibor Vamos were two presidents and they were outstanding, but we had to have a large number of people in Senior IFAC positions from those countries it sort of tailed off.

Rob:

Was there some sort of formal relationship between IFAC and IEEE?

Brian:

No, but the American AACC, they are the American NMO and of course IEEE linked in with them, so that’s some sort of through an intermediary there was.

Rob:

And I guess the other thing that occurs to me, you have some very productive collaborations in Europe.

Brian:

Yes.

Rob:

Were they as a result of you spending a lot of time with IFAC, or was it before that or nothing to do with that?

Brian:

No, If I look at the three greatest collaborators that I think I’ve had in Europe and I’ve had many and some have been very substantial but the three greatest were Mohamed Mansour at Swiss Federal Institute of Technology in Zurich, Michel Gevers at Catholic University at Louvain-la-Neuve, that’s the French one and Manfred Deistler at the Technical University of Vienna. And all three arose independently of IFAC and probably started before IFAC. That having been said, the IFAC secretariat is located 20 kms or something outside of Vienna, so I was duty bound to go to the IFAC secretariat once a year and someone paid my fare, so it was fortuitous and it helped sustain my relation with Manfred, and on occasions I would use such a trip and go and visit Michel. You can do multiple things on the one trip, but with Manfred it worked particularly well.

Rob:

Ok. That’s very good. I guess in reasonably recent times you became president of the Australian Academy of Science which is the equivalent, I guess in a sense to the National Academy of Science in the United States, which again, if I am right you were the first engineer to be president and that must have been also a fairly interesting period of your life cause these things can be political and complicated.

Brian:

Yes. Well there may have been other people with engineering qualifications, I am not sure who, being a biologist or something or other, but yes I was the first sort of genuine engineer. I don’t think that created any problems at all, which I am pleased about and I mean it’s a case of certain maturity you might say of people in that academy. Of course in such a role there are challenges and one of the challenges that I thought we faced was general low interest by the federal government in funding of R&D and a sense of disenchantment in the government with universities. Also the minister who looked after higher education was not particularly close to the Prime Minister and the way government in Australia works, as I’m sure you know, is that some ministers have the ear of the Prime Minister more than others and that’s useful if you can convince this Minister of the merits of the particular course of action. So for me the biggest issue when I was President was trying to do something about this. And fortunately as a president of the academy it is a kind of an entrée and it’s fairly easy to meet ministers and I had been previously a member of the Prime Minister’s Science Council and I knew a lot of these people. And as president of the academy I was an ex-officio member of the Prime Minister’s Science Council and that was a great platform. Now, there were a lot of bureaucrats who shared the view that the scientific community had about the government’s attitude, there were some ministers who were sympathetic and so it was a task of building a consensus to try to do something about that which really happened through the Prime Minister’s Science Council. And the funding for the Australian Research Council was doubled in about 2001, I think, and NICTA (National ICT Australia) money was found.

Rob:

Yes, so they were significant achievements

Brian:

They were not due to me, of course not, but I was certainly part of that process and worked very hard on it.

NICTA

Rob:

And you also founded and were the founding CEO of NICTA which is a very large research centre in the ICT area that also supported signal processing, control, so that was quite an important development for control and signal processing research in Australia, that’s for sure.

Brian:

Yes, yes.

Rob:

You know, have you got any reflections on being founding CEO of NICTA.

Brian:

The prime reason why it had to exist I think, is that government labs in the ICT area were not doing a great job and insofar as you can make international comparisons the universities weren’t doing a great job in computer science area, better in communications but computer science was pretty bad and there were tensions among the Ministers and differing views, with some of them thinking this was serious, like the defence people don’t like this at all and the Education Minister thinks everything is fine and he doesn’t want his portfolio interfered with. So anyway the government decided that they would start a new national laboratory with distributed campuses in ICT that it would interpenetrate the universities where there was some excellence in the universities to help get the organisation going faster, and to help drive the educational side of the enterprise. And so in contrast to almost all other government labs in this country it was set up more on the model of INRIA or CNRS in France or the Fraunhofer Institute or that sort of thing in Germany, and I was part of the process that helped the Government come to the view that there should be such an organisation. They ran a competition as you know, there was a consortium of universities at one time, your University was a member then it dropped out and later joined fortunately, and we put together a bid and won the bid. It involved me being CEO; I don’t think it would have been successful otherwise because I was kind of a trusted figure but I put a limit on it of a year and I had to do it in Sydney and I found it very difficult to operate out of another city which was not my own. I left my wife back in Canberra, and I never want to do that again. I have great respect for politicians who can travel to Canberra week after week from Perth or Adelaide, it must be horrible, any way that’s NICTA.

Other Collaborations

Rob:

Brian we talked a bit about your engagement with Boeing and industrial stuff from Stanford and Boeing and many other cases that you didn’t talk about and also your government engagements, but do you want to talk a little bit more about, cause I know you’ve had many industrial engagements with Cochlear and many other companies?

Brian:

Perhaps I can mention two of the most interesting, so Cochlear Corporation is the world’s major supplier of implanted bionic ears. They go in people who are maybe born deaf or have some serious problems. It is not one that is addressable with conventional hearing aids. So the company supplies, I’m guessing at the moment, 70% or so of the world’s (market).

Rob:

I think they’ve done 300,000 worldwide now.

Brian:

And it is worldwide and so it’s an enterprise doing wonderful things and it does them also very well and the people associated with them, like surgeons and nurses and the patients in the main, they are all very nice people. And I met many of these people, we’ve had board meetings abroad, I don’t think I’ve ever went to Africa but I certainly went to South America, North America, Europe and Asia with some board meetings and this was a wonderful experience which I did for ten years, and I was with the company from the day it listed. I’d had an earlier involvement with predecessors of the company for about seven years prior to that, but it listed on the Australian stock exchange and then I went ten years which is kind of the use-by date.

Rob:

I hadn’t realised it was that long, it’s quite an involvement.

Brian:

So the other one I was going to mention was called Pangea, and a group of people had the vision that Australia would be a wonderful site to provide a repository for high level nuclear waste. And that’s because of the geological stability of Australia, the fact that these highly geologically stable areas are not economically productive, you know there’s desert, there’s virtually no towns near some of them, there’s huge areas of Australia like this and part of the vision was that there could be safety in the transportation and that Australia had a stable political system. It was also possible once you’ve put this stuff in there you could effectively seal it all up, so it would be incredibly difficult to get it out again and there was, at least at that time, about seventy thousand nuclear warheads in eastern Europe that were tempting targets to underpaid former Soviet army officers and you know, nasty individuals of all shapes and sizes. So whatever you think of nuclear power this seemed to me to be a worthwhile thing to do and the company was headquartered in Switzerland but it set up an advisory scientific advisory board, US, Japan, UK, Swiss, and I was one of the several Australians. But the Australian Government at the time saw too much electoral liability in even engaging in a discussion, which is a shame. I understand governments take these decisions but I still think it is a shame because I think the world would have been a better place.

Rob:

Yeah, It’s intriguing. It might be a good time to talk about some of the things you think about the future of control. You talked about NICTA with computer scientists and systems and control engineers and bringing together the ICT space. What do you think about the impact of Computer Science on control systems, in various times people have talked about computer science will sort of take over the control domain or ubiquitous sensing and all these sort of things. It doesn’t seem to have happened and there seems to be a lot of exciting new things to do in control theory but certainly computing is playing an increasing role I think, have you got any thoughts about this?

Brian:

Yes, I think there are lots of points of contact between computer science and control, you know machine learning, the notion of doing things on a decentralised basis, event based control, this sort of a semi computer science notion. On the other hand it’s pretty rare for you to hear a computer scientist using the word feedback or using notions of dynamic systems to characterise what’s going on. A computer scientist is more interested to know whether an algorithm will terminate than whether the algorithm will give you a series of values that converge exponentially fast to wherever. I think it is the case of once you get into serious applications you generally need a collection of skills that you’ll almost never find in the one person and it will be common in applications whether they’re in health or control of water systems or operation of automatic freeways or automation with optimisation of traffic light systems and managing all this type of stuff. One is going to need a mixture of control people and communications people and computer science people, no question. I don’t think there is any danger of computer science taking over control, or maybe danger is the wrong word, but I think that’s unlikely to happen. I think that control future (is) in part driven by potential areas of application, and you’ve got governments worrying about health and ageing and environment and the environment thing includes smart grids and transportation, automated highways and that sort of stuff; and then you’ve got companies worrying about collecting masses of data so that they can better market, which is I guess a form of feedback and companies thinking about putting little sensors into everything you know, whether it is a bridge or car keys or what have you, so it’s ubiquitous but often hand in glove with communications and computer science talking about control.

Reflections on Control

Rob:

Yes, no I think it is an exciting era for control too and it seems to me that the huge body of knowledge that has been built up over the last fifty years about feedback and the nature of feedback is the great gift that control brings to many disciplines. My feeling is that it is not fully harnessed yet, that gift.

Brian:

People outside control I think find it fairly easy to grasp the concept of feedback when you can describe how a temperature control system works in a room but when you move onto the notion that feedback can maybe destabilise, oscillations and all that sort of thing in other words the dynamic side, it becomes intellectually far, far more complicated.

Rob:

I think that’s right.

Brian:

The control people of course are across that sort of thing.

Rob:

Perhaps computer science course should have some control courses in them too.

[laughs]

Advice for Young People

Rob:

It’s been a fascinating discussion Brian. Have you got any advice you would offer young researchers, I suppose particularly control researchers, but young people venturing into research careers?

Brian:

I think whatever your discipline is, it’s great to be passionate about it and its great to be good at it and then if you can try and associate with the very best people you’ll learn faster and may be become a good person yourself.

Rob:

Ok, that sounds like very wise advice. Thank you very much Brian.

Brian:

You’re welcome.