Å®ÉúСÊÓƵ

Computer scientists in Australia have developed a computerised air traffic
controller which can juggle the timing of arrivals and departures at airports
to minimise delays. Its designers hope to sell the concept to a number of
European airports suffering delays when the number of flights exceeds the
capacity of their runways.

The designers believe that reductions in errors of timing, more sophisticated
scheduling and fewer ‘lost slots’ in the sequence due to flight changes
will produce a 7 to 10 per cent improvement in an airport’s capacity. At
Sydney’s congested Kingsford-Smith Airport, even a small rise in capacity
should substantially cut landing and takeoff delays which, early this year,
lasted up to an hour.

The system is known as OASIS, for optimal aircraft sequencing using
intelligent systems. It was developed by the Australian Artificial Intelligence
Institute (AAII) in Melbourne and the Australian Civil Aviation Authority,
and will be tested at Kingsford-Smith Airport in October.

‘What we are tryng to do with OASIS is reduce the need for airborne
delays and holding of aircraft over airports,’ says Andrew Lucas, AAII’s
technical manager. Instead, OASIS will direct aircraft to start later or
slow down en route, which saves fuel and is less annoying to passengers.

AAII’s research at Kingsford-Smith Airport found that flow managers,
who con trol the sequence of arrivals and takeoffs, experience very busy
periods during which they can cope only by adopting a ‘first come, first
served’ policy.

OASIS uses radar data and information about aircraft performance and
individual airline practices to generate the sequence of arrivals and departures
that will cause the least delay.

Systems similar to OASIS are already operating at Frankfurt and at Paris’s
Orly Airport, but they are designed for only one runway. OASIS can handle
two or more runways, including runways which are used for both arrivals
and departures.

The OASIS computer program is divided into a collection of so-called
reasoning agents, each dealing with a different task. Each agent has four
components: a database containing the system’s current beliefs about the
world; a set of current goals; a library of plans describing what actions
need to be taken to reach the given goals; and an agenda or set of those
plans which have been chosen for execution.

Different agents concentrate on individual aircraft in the region or
on coordinating groups of aircraft. A sequencer agent, for example, carries
out a search of all possible combinations of aircraft on all available runways
to generate the most cost-effective sequence. A coordinator agent oversees
the tasks of the other agents. The system is overseen by the flow manager,
who can enter new procedures through a graphical display screen.

‘It’s a goal-directed system,’ says Lucas, ‘but if the beliefs about
what is going to happen begin to change, then it can suspend its reasoning
and select new goals.’

Computerised air traffic control systems, he says, have a poor image
with air traffic controllers, but OASIS was developed with the close cooperation
of controllers. If successful in Sydney, it will be extended to cover all
major airports in Australia.