Вы находитесь на странице: 1из 4

Proceedings of ECTI-CON 2008

Low Cost Sensory Based Intelligent Vehicle


Algorithm
Salilthip Vatcharadejsakul, Sanchai Thumrungsaisakul, Worameth Tachaplalert,
Kroeksak Tangmankhong, and Surapont Toomnark
Computer Engineering Department, King Monkut’s University of Technology Thonburi,
Bangkok, Thailand.
E-mail: nooice@gmail.com, nooyocpe@gmail.com, wora.pop@gmail.com,
kroeksak@cpe.kmutt.ac.th, surapont@cpe.kmutt.ac.th

Abstract-Autonomous system have been interesting topic in react against the surrounding environment. These vehicles are
the world of robotics, in order to replace human being. becoming widely used in both military and civilian sectors. For being
Intelligent vehicle is one of the popular autonomous system that autonomous, the vehicle needs sensor systems, computation and
gives many advantage for this world, not only in military but also
civilian. Route control and obstacle avoidance are now control system, and navigation system. Thus, the system must be
considered as major problems in autonomous system. In this real-time and embedded.
paper, new methods of Routing and Avoidance Algorithm for
Intelligent Vehicle are proposed. It is about using an artificial
C. Navigation System
intelligence algorithm together with avoidance system for Global Positioning System (GPS) is used for identifying the
navigating those vehicles. With reference to Global Positioning position on earth. At present, the system consists of 24
System (GPS), the robot will recognize its own position on earth satellites [4,5], which controlled by the on-grounded station.
and its destination by approximation. To improve the GPS can be used anywhere that the signal is reachable. GPS
approximating ability of the position defined, sensor system is elements divided into 3 segments. There are Space Segment,
used as additional reference. The computation process is used for User Segment, and Control Segment. In order to define the
calculating the route for the vehicle to move. If there is any
obstructer on the route, avoidance algorithm will be used for
position, we need the signal at least from 4 GPS satellites. Due
avoiding the collision. Real-time tracking and sensor-based to the error of GPS data, GPS is used as reference data of the
system are main sensory devices for testing avoidance algorithm. position. This system use only GPS receiver which is the
equipment for User segment. This equipment is used to
I. INTRODUCTION determine its location, speed, direction, and time.
Intelligent robots are used in many applications, especially D. Avoidance System
for autonomous vehicle. Without human control, autonomous One of the most essential capabilities of the intelligent
system has its own ability to run its tasks. Autonomous vehicle is obstrucle avoidance. The main objective of
vehicle must have ability to navigate itself to the specified avoidance is to detect obstacles and avoid them. To avoid
destination without any collision. While moving, there may be collision, variety of sensors are tend to be used to moves the
obstructers on the route that are not considered on the map. vehicle safely through uncertain area. Three ultrasonic sensors
Therefore the system must have its own avoidance algorithm are used to detect object at the front, left and right. Compass
for its making decision process. During the research process, Module is deployed to find the direction vectors of the car.
we use prototyped car instead of the real vehicle. Rotary encoders measure distance while vehicle is moving. To
make the system low cost, we considered to use only
A. Artificial Intelligence ultrasonic sensors as equipments for obstructer detection
John McCarthy, who is called the father of Artificial instead of using more expensive sensors.
Intelligence (AI) gives AI definition that “It is the science and
engineering of making intelligent machines, especially II. PROTOTYPE SYSTEM OVERVIEW
intelligent computer programs. It is related to the similar task
of using computers to understand human intelligence, but AI
does not have to confine itself to methods that are biologically
observable [1]”. This proposed algorithm use Hill-Climbling
heuristic search algorithm for choosing the shortest path to the
destination. This search algorithm is less random than
previous research using genetic search algorithm [2].
B. Autonomous System
The vehicle that can drive by itself without human control is called
Autonomous vehicle. This type of vehicle needs its own ability to
proceed the decision, senses, navigation and control [3] to learn and Figure 1, The prototype system Block Diagram

978-1-4244-2101-5/08/$25.00 ©2008 IEEE

653
From the figure above, the prototype system that was
selected for the navigation algorithm’s development and
verification is designed to be consisted of 3 modules,
Navigation Module, Route Control Module, and Avoidance
and Driving Module.
Navigation module uses GPS device to get earth position to
determine current point and destination. Route control module
manipulates route planning process of the car and guarantees
the car position on map. Avoidance and driving module’s
function is to move the car and avoiding the obstructers.
Avoidance system is main function for Avoidance and driving
module. Figure 3, Generating direct path from current point to destination
The prototyped car is built to be used for the project
experiment. This car consists of 2 DC Motors, ET-SMCC
V2.0., the hybrid motor control circuit, FPGA, and
ATmega128. For the sensor system, currently there are three
Ultrasonic Sensors, two Encoders, and one Digital Compass.
The car is installed with three GPS modules for the current
system experiment. The car will communicate with PC by
using wireless network, via Zigbee communication protocol.

Figure 4, Finding the route according to the real roadmap data

GPS system used for this project is in User segment. GPS


receiver will continuously compute the current position while
the car is moving. There are many additional packets can be
captured by the GPS module, for example, the height from the
ground, but we only consider latitude and longitude data. GPS
data have errors, which cannot be avoided, but can be reduced
by continuous experiment and by finding average values from
Figure 2, The prototype car
many GPS modules at that same time.
III. PROPOSED ALGORITHM For this algorithm, mapping process also use sensor system
to determine the distance and simulate temporary map to be
The methods used for the algorithm are using GPS module
compared with the map data from GPS. This procedure can
as equipment for defining position on earth and make the car
reduce the GPS error problem.
moves without any crash. But weakness of GPS system is its
method of using the received signal from satellites. Some
environments, for example in urban area, may cause the
reflection of signals which may cause error in positioning.
Therefore sensor system is used for guarantee the position on
the real map and it is also used for avoidance system. The
proposed algorithm is divided into two main modules which
are Routing Algorithm and Avoidance algorithm. Routing
Algorithm is in the calculation process of finding the path for
the car to move. Avoidance algorithm is the motion control
algorithm that is sensor-based system.
A. Routing algorithm
Routing algorithm use GPS for positioning process, both for
defining current position and destination. After get those two
positions, the system will start computes the path to move the
car. First, the system will generate the direct path to the
destination in terms of straight line. After that, the system will
find the route according to roadmap in the database by using Figure 5, GPS data received and the filtered GPS data
straight line as standard path.

654
B. Avoidance algorithm lower than 0.5 meter. The procedure of the avoidance
The algorithm begins with detecting obstructer that is found algorithm is shown in figure 7.
in front of the car. This avoidance algorithm is divided into
two parts. For the first part, the data from sensors are
considered as the object detector around the car. This can help
the car avoid any crash and stay in appropriate distance. The
sensors used in this part are SRF04 Ultrasonic sensors,
CMPS03 Digital Compass Module, and Encoders. Three
Ultrasonic sensors are used to determine the distance between
the car and the objects. Another part is to measure the distance
that the car moves and its heading direction. Encoders are
used to count the distance that the can have moved. Compass
Module is used to find the headed direction of the car.

Figure 6, Avoidance procedure

From the figure above, step A, the system will


automatically generate path for the car to move by using the Figure 7, Flowchart of Avoidance Algorithm
route data in the database. Step B, as the car finds obstructer
on its path, Obstructer Avoidance Algorithm will corrupt to There are three cases of avoidance use in this algorithm.
avoid obstructer, as shown in step C. The car will continue First case is the object is in front of the car, the car will avoid
running until it cannot detect any objects as shown in step D. to the right. Second case is the object is in front of the car and
Step E, the car will recomputed again in order to navigate in right front or right front of the car, the car will avoid to
itself back to its path. another direction. The last case is the object blocks the road,
FPGA, the high-speed system-on-chip controller with the car will move backward, and then it will map the blocked
parallel computing ability [7] is the main component of this path on the temporary map, and recomputed its route again.
avoidance system. To detect obstructer, this Avoidance system
use Ultrasonic sensor. When the car moves, ultrasonic sensor IV. EXPERIMENT
will continuously spread out its beams for detecting objects on
its route. If there is object on the route, avoidance process will A. GPS Experiment
be triggered. Compass module uses the method of comparing In the project experimenting area, there are nothing can be
the data received with the earth's magnetic field in order to reflected by the GPS signal, but it also gives some errors. After few
reports heading angle using north direction as reference. experiments, the result shown that GPS errors will affects the
Using three ultrasonic sensors do not cause errors of the accuracy of the system. Up to this date it is used as the reference data
overlapped signal because of the beam patterns of SRF-05 of the position. Since the experimenting area is on the top of the
ultrasonic sensor is not too wide and we only consider the data building which the satellite signals will not be reflected, so the system
received when the distance between the object and the car is will get so little effect from the error of GPS. The area consists of 15

655
reference points which are used for the system’s accuracy V. CONCLUSION
experiment.
This research attempted to design the new algorithms for the
autonomous vehicle. Our framework consists of two major
algorithms, Routing algorithm and Avoidance algorithm. The system
is experimented using a prototype car. As the experiment result, the
car has ability to learn, compute, control, and navigates by itself in
specific area with undeclared obstructer. However, the research is
currently conducted in a controlled environment. For the future
implementation, we have planned to make the system able to handle
with many other additional areas. An application of autonomous
navigation, in environments where there are many unexpected
obstacles, may exist if we want to implement this algorithm in the
real system. The major characteristic of this algorithm is how to
make the car move to its direction without any crash even there are
unexpected obstructers.
ACKNOWLEDGMENT
Figure 8, Difference between expected GPS data and the real data captured
This research has been supported by faculties of Computer
B. Avoidance and Sensor system Experiment Engineering Department, KMUTT. I would like to thank Asst.
Test platform is simulated to test the abilities of avoidance system. Prof. Surapont Toomnak and Mr. Kreoksak Tangmankhong
The avoidance system uses our proposed avoidance algorithm. The for every supports and recommendations.
result is the car can move successfully to the destination, it can avoid
obstructers and get back into its route. There are some errors during REFERENCES
the experiment, for example sometime the back of the car stuck to the
[1] John McCarthy, WHAT IS ARTIFICIAL INTELLIGENCE?
corner of the obstructer while it try to get back to the route. http://www-formal.stanford.edu/jmc/whatisai/whatisai.html
[2] N. Baba, N. Kubota, Collision avoidance planning of a robot
manipulator by using genetic algorithm. A consideration for the problem
in which moving obstacles and/or several robots are included in the
workspace, IEEE World Congress on Computational Intelligence.,
Proceedings of the First IEEE Conference on 27-29 June 1994, vol.2,
pp.714 - 719, 1994.
[3] Ljubo Vlacic, Michel Parent and Fumio Harashima, Intelligent vehicle
technologies : theory and applications, Butterworth, UK, 2001.
[4] Michael Kennedy, The global positioning system and GIS : an
introduction, Taylor, USA, 2002.
[5] James Bao-Yen Tsui, Fundamentals of global positioning system
receivers: a software approach, John Wiley & sons, 2000.
[6] Y.K. Kwag, J.W. Kang, Obstacle awareness and collision avoidance
radar sensor system for low-altitude flying smart UAV, Digital Avionics
Systems Conference 2004 (DSAC04), Vol.2, pp.12.D.2 - 121-10, 2004.
[7] FPGA Discovery-III XC2S300 Board Manual
http://ailogictechnology.com

Figure 9, The avoidance experiment

656

Вам также может понравиться