Although the steps are clear, translation into real world
application is challenging. Each input is measured from a
sensor and only when a certain condition is met, this is to
estimate if there is an accident in certain zone the system
must know when to check for an accident in a zone. A
system that does not know when to start the calculations
will not be able to tell the difference between congestion
and a normal flow in real time or will gather too much
unneeded and redundant data that will cause overhead to
the system and distract it from taking reliable decision. The
system must rely on some known data about what normal
flow means and what congested road should look like. This
trusted data is gathered over time and with through
analysis is made useful to describe the road’s behavior.
After collecting enough data, the fuzzy system then will
start comparing the usual pattern of traffic flow with the
current pattern of the road at certain periods that are
triggered by flags. Those flags are the Cross Ratio, Gap
Filling Time and vehicles’ speed. Those three elements
play important role is deciding when to activate the system,
wither there’s an accident or not and where is the accident
exactly located, prospectively. Then later on, an action will
be made by the traffic system to respond to those states.
Because streets are considered vague environment that are
hard to monitor, fuzzy logic is a perfect way to extract
useful information from inaccurate data collected from the
sensors. Fuzzy logic uses Membership functions and fuzzy
variables to map collected data to more reliable terms later
on making decisions fast and reliable at the same time.