Page 13 - The Keble Review 2016

Stephen Cameron
30 Years a Fellow at Keble
I became interested in robotics while pursuing a PhD in Artificial Intelligence at the University of Edinburgh. After that I was invited by McDonnell-Douglas to do some post-doctoral work in St Louis, Missouri, which exposed me and and my wife, Frances, to that strange country with its beautiful countryside. However after almost two years the cold beer and lack of pavements started to get on our nerves, and on 1st September 1986
I started at Keble, first as a Research Fellow and then as a
tfull Tutorial Fellow in 1988.
hings were different then. Computers were expensive enough that people queued up to use them; e-mail worked across the country but no further; and the internet wasn’t
even a twinkle in Tim Berners-Lee’s eye. After a few years we started to get PCs into Keble – some into the Bursary and a couple for use by the JCR and MCR. A research student and myself spent a weekend wiring up the first crude computer network. Keble had been the first College to appoint a full-time tutor in Computer Science, and would be the first to start adding internet wiring into student rooms, but even that would have to wait for the turn of the century.
Meanwhile I was engaging in research into the use of industrial robots; other types barely existed then. Specifically, we wanted
to be sure that if an autonomous robot decided to move in a particular way that it wasn’t going to hit anything. This is an obvious problem for a human to solve, but a very difficult one
for a computer. There is also the problem that robots don’t go exactly where you expect, and we had to take such uncertainties into account. Gradually we realised that the answer was to plan at multiple levels of abstraction – a pattern that we still use today. It’s a bit like deciding how to walk across Oxford by deciding which roads to follow first, and where to cross those roads later.
During the 1990s British interest in industrial robotics diminished, and our focus moved slightly. The ideas we had developed for robotics also proved useful in the burgeoning computer games market; indeed, they are probably still in many existing games. And we also had an opportunity to try out our ideas in a non- industrial setting with the development of the Robot Sheepdog,
which (despite its name) was used to demonstrate some ideas from the theory of animal behaviour by building a small robot that could herd ducks in a controlled manner.
A big change since the turn of the century has been the ability
to try out new robot designs quickly, with the dramatic drop in cost of components, computing, batteries and sensors – much
of this due to the growth in the mobile phone industry. We took
a group of small ‘sheep’ robots to the Royal Society Summer Exhibition in 2001, and have used small wheeled, legged or flying robots for public demonstration purposes ever since. Our work on legged locomotion has led us to attend several of the RoboCup robot tournaments, and nowadays we regularly run sessions with school parties based around a series of robot races in order to help enthuse children about STEM subjects.
Over the last ten years we have continued to find uses for ideas from robotics in other areas, such as pharmaceutical drug design and medical imaging. We have worked with small robot helicopters for some time, as these have many potential uses in robotic search and rescue and in inspection. One big question is how we can develop autonomous devices so that the public and the regulators trust us to use them; that has led me into the whole area of drafting of regulations, and I am now the Chair of the British Standards Institute committee that deals with standards for robots.
Keble had been the first College to appoint a full-time tutor in Computer Science, and would be the first to start adding internet
wiring into student rooms
Computer Science teaching in Oxford was restricted to the joint degrees with Mathematics or Engineering for many years, with a full degree only appearing in the late ‘90s. It has taken some time for would-be applicants to understand how interesting and useful the subject is, but over the last five years the number of applications has more than doubled, and the College has now taken on Standa Živný as the second Fellow in the subject. The future of Computer Science at Keble is surely secure.
Professor Stephen Cameron is a Fellow and Tutor in Computer Science. His general area of interest is in spatial reasoning, which includes the planning of tasks and motions for robot vehicles and manipulators, the use of geometric models, and the scheduling of fleets of robots. The College celebrated his thirty years at Keble at the Maths and Computing Reunion Dinner on Saturday 21 May 2016.
13