Robot wars
Author: By Samantha Keen
Date: 24 Apr 2002
Words: 755
Publication: Sydney Morning Herald
Section: Supplement
Page: 7
 
We'd all like to see Hal come to life, but the realities of artificial
intelligence and its benefits are far more pedestrian.  Australia is
the home of the world-champion robot soccer team. Robots developed by
the School of Computer Science and Engineering at the University of
New South Wales proved Australia's worth in developing artificial
intelligence when they beat Carnegie Mellon University of Pittsburgh,
9-2.

"We were pretty pleased about that because they have a very
respectable team of researchers in AI," says Professor Claude Summut,
School of Computer Science and Engineering at UNSW.

When most people think of AI research they think of Hal, the computer
in 2001: A Space Odyssey. A computer with intelligence like Hal might
be the general direction in which scientists are heading, but in the
shorter term AI research has less grandiose benefits for you and
me. For instance, voice-recognition software used by telephone
companies, face-recognition techniques used by police and Internet
search engines.

"AI seems to be all about the things we don't know how to do, but once
we work it out then it is no longer classified as AI," says Dr Yusaf
Pisan, department of computing, Macquarie Univers-ity. "For example,
it used to be that playing really good chess was something we thought
an AI computer could do, but now we have computers who can beat world
champions in chess and they are not considered AI." Pisan uses game
technology for a large part of the department's AI research. Many of
today's games have input from academics working in areas such as AI -
researchers and academics tend to innovate the major developments that
are in turn polished and upgraded by the games companies. University
academics have the time to develop ideas rather than catering to the
time demands of the market if they worked in large corporations.

Pisan and his team are working on making autonomous characters and
monsters in computer games such as Half-Life and Unreal Tournament.

Australian postgraduate studies in AI are booming. UNSW is one of the leaders in the field with about 30 PhD students working in AI, many with robotics. UNSW also specialises in computer vision, which means developing ways for the robots to see that soccer ball and judging how to pick it up or kick it. 

It also works in the areas of knowledge representation, which means
finding methods for a computer to reason with complex concepts that
humans often take for granted.

For example, when we know we have to go to the airport we
automatically know we need to phone a taxi and to call early enough to
give the taxi warning. It is challenging to teach a computer to know
these kinds of things.

UNSW is also working in machine learning with the aim of giving
machines the ability to improve their own performance through, say,
trial and error. ANU, Sydney University and Wollongong University are
involved with UNSW in the Smart Internet Technology Creative Research
Centre, jointly funded by the universities, government and corporate
sponsors.

UTS emphasises corporate relevance in its study of AI, having recently
won a grant for a program using AI in the e-market setting. The group,
including about 10 PhD students, is developing multi-agent systems for
data mining or intelligent agents that can track through large blocks
of information and work together to give the user a small and relevant
choice of research. The agents would also be used for buying and
selling in the market setting.

"Basically, a piece of software is an agent that behaves
autonomously," says Professor Simeon Simoff, senior lecturer, Faculty
of Information Technol-ogy, UTS. He says negotiation skills such as
these are difficult and complex.

"Take the Middle-East, what is negotiation for them? If we come up
with something really useful about negotiation ..."

The group wants to have a product available for market use by the end
of the year, completing the project at the end of 2004.

Professor Simoff says the short-term projects and corp-orate emphasis
are part of the UTS approach, as opposed to Sydney University, where
he has studied: "Sydney University is more theoretical and not as much
related to industry needs. They are happy to develop a project that
might take 10 or 20 years to hit the market."