Innovative approaches to
handling uncertainty in air traffic management
The 21st century finds Europe facing a number of
remarkable changes, many of which involve safety critical complex operations
that have to deal with uncertainty, and air traffic is one of them. The
management and control of these operations are undergoing a natural trend of
becoming more and more distributed, while at the same time their safety
criticality for human society tends to increase. Nonetheless, it is human beings
who continue to bear the responsibility for safety, and hence they need ways to
manage the uncertainties in a predictably safe way.
Within the European Commission project HYBRIDGE, some
50 system theorists and mathematicians from 6 universities (Cambridge, Twente, L’Aquila, NTUA, Brescia and Patras) and from 3
research institutes (NLR, INRIA and CENA) have studied the development of
innovative approaches to handling uncertainty in complex safety critical
operations. These theorists have brought together state of the art mathematical
and computer science approaches to handling uncertainty in automation, finance,
robotics and transport. In collaboration with experts from AEAT, BAES, EEC and
NLR these approaches have been elaborated for air traffic. The very reason for
selecting air traffic is that the interactions between the many agents (humans
and systems) are so highly distributed in comparison with other safety critical
operations.
Figure
1. Air
traffic management compared with other safety critical operations in terms of
potential number of fatalities per accident and the level of safety critical
interactions between human and automata agents
In mathematical terms the innovative approaches to
handling uncertainty are based on esoteric but powerful studies of stochastic
hybrid systems, where ‘hybrid’ refers to the integration of computer science
and control theory. The resulting reports and papers are available on the project web site.
Because of these powerful underlying concept studies, the basic results
obtained add significantly to what is known in safety critical industries. And
for air traffic in particular, these results have been elaborated into the
following three innovative approaches:
1. Managing the combinatorial nature
of non-nominal safety critical conditions. The
application of mathematics towards handling combinatorial situations is well
known. The key novelty however is the recognition that the larger part of the
combinatorial safety critical conditions have to do with Situation Awareness
(SA) differences between multiple agents (humans and systems), and the
systematic modelling and evaluation of the resulting SA and SA error
propagation. This innovative approach is illustrated for two well known safety
issues in air traffic: i) Runway incursions, and ii)
Level busts.
2. Accident Risk Assessment of air
traffic designs by Monte Carlo simulation. The key
behind this development is an innovative way in modelling non-nominal events
and responses by pilots, controllers and systems, which connects the resulting
simulation code unambiguously to a stochastic and discrete event mathematical
model. This connection allows using mathematically powerful tools to speed up
Monte Carlo simulations by 6 to 8 orders of magnitude. Such a large speed up
factor makes it possible to estimate the probability of a collision between
aircraft through Monte Carlo simulations. The effectiveness of this innovative
approach is demonstrated for an Airborne Separation Assistance System based
en-route operation.
3. Monte Carlo simulation based
optimisation of air traffic management. The
optimisation problems that arise in air traffic management are formidable. They
involve multiple and complementary objectives in terms of safety, economy and
environment, whereas there are major sources of uncertainty such as weather and
traffic demand. Recent advances in randomized optimization methods provide an
innovative approach in addressing such complex problems. Within the HYBRIDGE project,
randomized optimization methods have been further developed and initially
applied to collaborative decision support problems that arise in en-route
conflict resolution and arrival sequencing.
Invitation
On 9th February 2005, the HYBRIDGE consortium
will present and demonstrate these innovative approaches to the ATM community.
Interested researchers are cordially invited to attend this workshop, which
will be held at the Eurocontrol Experimental Centre in Brétigny-sur-Orge,
France. If you would like to attend or receive further information please send an email to Richard
Irvine or to Henk Blom,
the HYBRIDGE co-ordinator.