20-09-2012, 03:05 PM
Adaptive Car Safety System
[attachment=34170]
Abstract
Car accident is one of the major causes of death in
many countries. Many researchers have attempted to design and
develop techniques to increase car safety in the past recent years.
In spite of all the efforts, it is still challenging to design a system
adaptive to the driver rather than the automotive characteristics. In
this paper, the adaptive car safety system is explained which
attempts to find a balance.
INTRODUCTION
N our twentieth century, transportation is one of the
greatest industries. Many experts are currently working
on different projects to increase the safety of automobiles
and decrease the number of accidents on the roads. Based
on the statistics, approximately 27,000 civilians are killed
in car accidents in some developing countries each year
[12-17]. In 2008, the number of accidents which result in
death increases by 15%. In other words, 80 people are
killed in car accident each day. Consequently, the next
generations of cars are improved so that the number of
accident decreases. Innovative ideas have emerged and
implemented in order to reduce the risk of car accidents.
During the past recent years, some alarm systems and
intelligent control apparatus have been designed and
developed in order to increase the safety of automobiles.
DESCRIPTION OF ADAPTIVE CAR SAFETY SYSTEM
As one can see in the system diagram (Fig. 1), the
preprocessing stage receives four inputs: the first two
inputs are velocities which were read through the timer
counter of the microcontroller. Acceleration in X direction
has been implemented by variable resistors and a heavy
weight attached to it and it is read by the analog to digital
converter (ADC) of the microcontroller. The deviation of
the weight results in the deviation of the variable resistor
and corresponds to the acceleration in X direction. The
same procedure is repeated for Y direction as well. The
next input (role angle) is estimated by using a shaft encoder
which has been attached to the role. In the preprocessing
stage, the information is changed in a way that can be read
by other components of the system. In order to determine
the threshold for maximum speed of a car, an artificial
neural network is utilized. Since the behavior of the system
changes according to the location of the car (city, town,
road, street, etc.), the weights of the network are loaded
corresponding to the location. Detection of the location is
determined by measuring the changes of the role angle,
acceleration, and GPS system coordinates.
DETECTION OF DANGEROUS POINTS USING GPS
In this project, a Garmin five-meter precision GPS has
been utilized [8]. X and Y coordinates of the current
location are provided to the controller by the GPS.
Therefore, it can easily refer to the location of the car
whenever it needs to make a decision. GPS has been
connected to the serial port of the micro-controller using
RTCM/TEXT protocol, and a speed of 4800 Megabits per
second. It also provides velocity, height above sea level
(altitude), and other parameters.
Here, GPS has two major benefits: firstly, coordinates
provided by GPS, together with changes of the role angle,
speed and acceleration help to distinguish between city
streets and roads outside city. Secondly, comparison
between dangerous points of the map and current
coordinates of the car is facilitated. If the car is close to
risky locations, speed of the car will be limited. System
loads the proper weights into the neural network using the
previous information and starts to control the speed of the
vehicle. Since the information about 3000 dangerous points
takes about 24 kilobytes, the information must be stored in
external memory outside the micro-controller.
ARTIFICIAL NEURAL NETWORK
Although computers have improved a lot during the past
decades, there are still many problems which are really
time-consuming and difficult for computers to solve. In
fact, computers can easily solve complex arithmetic
operations on very large numbers in a fraction of second.
However, computers can not work efficiently when dealing
with noisy input, or inputs taken from the surrounding
environment, parallel processing on huge amount of data.
Also they can not generalize a small pattern to a bigger
one. On the other hand, human brain or even an animal can
easily distinguish and perceive the environment. Artificial
neural network attempts to imitate human brains. It tries to
mimic the human brain behavior and experience the
environment.
CONCLUSION
The experimental results show that our implemented
system compared with other anti-collision and car safety
systems adapt with the car and the driver to control the
speed of the car. An adaptive MLP (multi-layer perceptron)
cooperates with other decision making modules to control
the speed of car by controlling the amount of gas which is
pumped into the engine.
[attachment=34170]
Abstract
Car accident is one of the major causes of death in
many countries. Many researchers have attempted to design and
develop techniques to increase car safety in the past recent years.
In spite of all the efforts, it is still challenging to design a system
adaptive to the driver rather than the automotive characteristics. In
this paper, the adaptive car safety system is explained which
attempts to find a balance.
INTRODUCTION
N our twentieth century, transportation is one of the
greatest industries. Many experts are currently working
on different projects to increase the safety of automobiles
and decrease the number of accidents on the roads. Based
on the statistics, approximately 27,000 civilians are killed
in car accidents in some developing countries each year
[12-17]. In 2008, the number of accidents which result in
death increases by 15%. In other words, 80 people are
killed in car accident each day. Consequently, the next
generations of cars are improved so that the number of
accident decreases. Innovative ideas have emerged and
implemented in order to reduce the risk of car accidents.
During the past recent years, some alarm systems and
intelligent control apparatus have been designed and
developed in order to increase the safety of automobiles.
DESCRIPTION OF ADAPTIVE CAR SAFETY SYSTEM
As one can see in the system diagram (Fig. 1), the
preprocessing stage receives four inputs: the first two
inputs are velocities which were read through the timer
counter of the microcontroller. Acceleration in X direction
has been implemented by variable resistors and a heavy
weight attached to it and it is read by the analog to digital
converter (ADC) of the microcontroller. The deviation of
the weight results in the deviation of the variable resistor
and corresponds to the acceleration in X direction. The
same procedure is repeated for Y direction as well. The
next input (role angle) is estimated by using a shaft encoder
which has been attached to the role. In the preprocessing
stage, the information is changed in a way that can be read
by other components of the system. In order to determine
the threshold for maximum speed of a car, an artificial
neural network is utilized. Since the behavior of the system
changes according to the location of the car (city, town,
road, street, etc.), the weights of the network are loaded
corresponding to the location. Detection of the location is
determined by measuring the changes of the role angle,
acceleration, and GPS system coordinates.
DETECTION OF DANGEROUS POINTS USING GPS
In this project, a Garmin five-meter precision GPS has
been utilized [8]. X and Y coordinates of the current
location are provided to the controller by the GPS.
Therefore, it can easily refer to the location of the car
whenever it needs to make a decision. GPS has been
connected to the serial port of the micro-controller using
RTCM/TEXT protocol, and a speed of 4800 Megabits per
second. It also provides velocity, height above sea level
(altitude), and other parameters.
Here, GPS has two major benefits: firstly, coordinates
provided by GPS, together with changes of the role angle,
speed and acceleration help to distinguish between city
streets and roads outside city. Secondly, comparison
between dangerous points of the map and current
coordinates of the car is facilitated. If the car is close to
risky locations, speed of the car will be limited. System
loads the proper weights into the neural network using the
previous information and starts to control the speed of the
vehicle. Since the information about 3000 dangerous points
takes about 24 kilobytes, the information must be stored in
external memory outside the micro-controller.
ARTIFICIAL NEURAL NETWORK
Although computers have improved a lot during the past
decades, there are still many problems which are really
time-consuming and difficult for computers to solve. In
fact, computers can easily solve complex arithmetic
operations on very large numbers in a fraction of second.
However, computers can not work efficiently when dealing
with noisy input, or inputs taken from the surrounding
environment, parallel processing on huge amount of data.
Also they can not generalize a small pattern to a bigger
one. On the other hand, human brain or even an animal can
easily distinguish and perceive the environment. Artificial
neural network attempts to imitate human brains. It tries to
mimic the human brain behavior and experience the
environment.
CONCLUSION
The experimental results show that our implemented
system compared with other anti-collision and car safety
systems adapt with the car and the driver to control the
speed of the car. An adaptive MLP (multi-layer perceptron)
cooperates with other decision making modules to control
the speed of car by controlling the amount of gas which is
pumped into the engine.