seminar code
08-08-2014, 03:28 PM
Fuzzy Control System For Controlling Traffic Lights
[attachment=66760]
Abstract
This paper introduces a flexible technique for the
control of traffic lights using fuzzy Mamdani type controller.
The real time feedback parameters, traffic density, and queue
length are obtained using image processing techniques viz.
partitioning and segmentation. The on and off timings for
green, red and amber lights are adjusted as per the actual road
conditions. The defuzzification of the combined results is done
using Height method
INTRODUCTION
Today, we are living in the world of automation. Micro
controllers control most things around us. The control of
traffic lights is a well-known area where this type of control
system is incorporated. But, the control is not flexible, based
on the condition of traffic at the crossing. Rather, the on and
off time periods are fixed for the red, green and amber lights.
At most, these durations are varied as per the time of the day,
the day of the week etc. There is no real-time adjustment of
the on/off times as per the traffic conditions on the crossing.
This paper proposes a method for introducing more
flexibility in the control of traffic lights using visual feedback
from images acquired by cameras, image segmentation and
fuzzy control techniques.
THE PRESENT DAY TECHNIQUE
The most commonly used technique for traffic light
control today is based on a micro controller, which controls
the four sets of traffic lights at the traffic island/ crossing.
This is ordered to work such that the traffic in only those
directions, which do not cross each other, is allowed to move
at any given time (i.e. green light is displayed in that
direction) while in all the other directions, the traffic is
forbidden to move (i.e. red light is displayed in that
direction). For example, in a country following the left hand
drive system (e.g. India) the traffi
THE PROPOSED TECHNIQUE
The technique described above is a typical nonflexible
control technique. The technique being proposed in this
paper uses the technique that a human (traffic policeman)
would employ in the same situation. The human operator can
observe the actual condition of traffic and change the timings
according to the actual condition of queues and rush in
different directions while still maintaining the fair rotation of
turns. In contrast a micro controller will keep the light in a
direction as ‘green’ till the allotted time span is over, even
though there may not be any vehicles waiting in that
direction, whereas there may be a long waiting queue in other
Matching Degrees for individual Rules and Inference:
To find the membership value of the antecedents, the
actual values of Queue length and traffic density achieved
by image processing described above are fitted to the
antecedent membership curves described in section B
above. The matching degrees for the six fuzzy rules (R2 to
R7) are found by finding the minima of the membership
values thus found.
CONCLUSION
The System proposed here is very flexible. The timing
adjustment ratios and the minimum and maximum time limits
can be altered, while keeping the structure of the rule set
unchanged. The feedback for the queue lengths and traffic
densities can be taken from images taken from cameras above
(a bird’s eye view). Image processing techniques can be
applied to separate the areas of interest (the concerned lanes),
and analyze them to find the attributes viz. the density and the
length of the queue. Presently, this is a proposed theoretical
paper and has only been simulated in a lab, with satisfactory
results. The definitions of the fuzzy sets of the antecedents
are also very easily changeable. This is a very promising
application of fuzzy logic in practical areas, and will be
[attachment=66760]
Abstract
This paper introduces a flexible technique for the
control of traffic lights using fuzzy Mamdani type controller.
The real time feedback parameters, traffic density, and queue
length are obtained using image processing techniques viz.
partitioning and segmentation. The on and off timings for
green, red and amber lights are adjusted as per the actual road
conditions. The defuzzification of the combined results is done
using Height method
INTRODUCTION
Today, we are living in the world of automation. Micro
controllers control most things around us. The control of
traffic lights is a well-known area where this type of control
system is incorporated. But, the control is not flexible, based
on the condition of traffic at the crossing. Rather, the on and
off time periods are fixed for the red, green and amber lights.
At most, these durations are varied as per the time of the day,
the day of the week etc. There is no real-time adjustment of
the on/off times as per the traffic conditions on the crossing.
This paper proposes a method for introducing more
flexibility in the control of traffic lights using visual feedback
from images acquired by cameras, image segmentation and
fuzzy control techniques.
THE PRESENT DAY TECHNIQUE
The most commonly used technique for traffic light
control today is based on a micro controller, which controls
the four sets of traffic lights at the traffic island/ crossing.
This is ordered to work such that the traffic in only those
directions, which do not cross each other, is allowed to move
at any given time (i.e. green light is displayed in that
direction) while in all the other directions, the traffic is
forbidden to move (i.e. red light is displayed in that
direction). For example, in a country following the left hand
drive system (e.g. India) the traffi
THE PROPOSED TECHNIQUE
The technique described above is a typical nonflexible
control technique. The technique being proposed in this
paper uses the technique that a human (traffic policeman)
would employ in the same situation. The human operator can
observe the actual condition of traffic and change the timings
according to the actual condition of queues and rush in
different directions while still maintaining the fair rotation of
turns. In contrast a micro controller will keep the light in a
direction as ‘green’ till the allotted time span is over, even
though there may not be any vehicles waiting in that
direction, whereas there may be a long waiting queue in other
Matching Degrees for individual Rules and Inference:
To find the membership value of the antecedents, the
actual values of Queue length and traffic density achieved
by image processing described above are fitted to the
antecedent membership curves described in section B
above. The matching degrees for the six fuzzy rules (R2 to
R7) are found by finding the minima of the membership
values thus found.
CONCLUSION
The System proposed here is very flexible. The timing
adjustment ratios and the minimum and maximum time limits
can be altered, while keeping the structure of the rule set
unchanged. The feedback for the queue lengths and traffic
densities can be taken from images taken from cameras above
(a bird’s eye view). Image processing techniques can be
applied to separate the areas of interest (the concerned lanes),
and analyze them to find the attributes viz. the density and the
length of the queue. Presently, this is a proposed theoretical
paper and has only been simulated in a lab, with satisfactory
results. The definitions of the fuzzy sets of the antecedents
are also very easily changeable. This is a very promising
application of fuzzy logic in practical areas, and will be