Docstoc

Customized Digital Road Map Building using Floating Car GPS Data

Document Sample
Customized Digital Road Map Building using Floating Car GPS Data Powered By Docstoc
					                                                                 (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                          Vol. 8, No. 3, 2010

           Customized Digital Road Map Building using
                     Floating Car GPS Data
                        G. Rajendran                                                                  Dr. M. Arthanari
         Assistant Professor of Computer Science,                                                        Director,
         Thiruvalluvar Government Arts College,                                       Bharathidasan School of Computer Applications,
          Rasipuram-637401, Tamilnadu, India                                               Ellispettai-638116, Tamilnadu, India
                guru.rajendran@yahoo.com                                                        arthanarimsvc@gmail.com


                                                              M. Sivakumar
                                                     Doctoral Research Scholar,
                                             Anna University, Coimbatore, Tamilnadu, India
                                                          sivala@gmail.com

Abstract—The vehicle tracking, navigation and road guidance                  physical features of an environment [2, 3, 4]. This information
applications are becoming more popular but the presently                     is taken from different positions along the path followed by a
available digital maps are not suitable for many such                        moving object. Once the map is obtained, it can be used to
applications. Among the drawbacks are the insufficient accuracy              improve the quality of the paths, to locate a target, to help
of road geometry and the delayed time in loading the unwanted                object recognition, to define expectations in the trajectory or to
data. Most of the commercial applications in vehicle tracking                replay the travelled path of a moving object.
require digital maps which have only roads and places of interest
whereas the currently available maps show all available roads                    Digital maps of required roads with good accuracy are
and places. A simplified map building process to construct                   needed in a number of commercial applications. But the
customized high-precision digital maps from floating car data                presently available digital maps are fully populated with dense
obtained from Global Positioning System (GPS) receivers is                   roads and other information, most of which are irrelevant to the
presented in this paper. The data collected from the GPS receiver            requirement. Because of the presence of these unwanted data,
fixed in a moving car are used to construct the customized digital           the loading time of the map in computer memory is also more
road maps. The approach consists of six successive steps:                    which results in slower execution of the application. Hence the
Collecting floating car data (FCD) for desired road segments in a            need for building customized digital road maps is essential and
log file; refining the log file; constructing the road segments using        such a map building process will also eliminate the expenses
the data present in refined log file; merging the segments which             involved in buying digital maps. This paper presents a
has negligible slope; refining the road intersections; and labeling
                                                                             customized digital road map building process which can be
the points of interest. The quality of outcome of the map making
                                                                             used to construct high quality maps of desired roads and
process is demonstrated by experimental results and the results
indicate that customized road maps of required routes with good              locations. The source code for the intermediate processing
accuracy can be built with the help of the proposed map making               steps is written in Matlab 7.6.
process.                                                                          Map building process has already been discussed by some
                                                                             of the researchers, but with limitations like complexity in map
    Keywords- digital map; global positioning system; floating car           building and insufficient accuracy. These limitations have been
data; road network
                                                                             addressed in this work. The remainder of this paper is
                                                                             organised as follows. Section 2 of this paper describes road
                        I.  INTRODUCTION                                     network model and the collection of floating car GPS data.
    Map is a total or partial depiction of the structure of the              Related work in this area is discussed in Section 3. In Section
earth (or sky) on a plane, such that each point on the map                   4, the map making process is discussed. The experimental
corresponds to an actual point on the earth (or in the sky).                 results are dealt in Section 5. The work is concluded and the
Digital maps are used to produce this depiction in electronic                possible improvements are discussed in Section 6.
form. The digital maps are usually represented as graphs,
where the nodes represent intersections and the edges are                      II.   THE ROAD NETWORK MODEL AND THE FLOATING CAR
unbroken road segments that connect the intersections [1].                                          DATA
Each segment has a unique identifier and additional associated
attributes, such as the number of shape points that approximate              A. The Road Network Model
its geometry roughly, the road type (e.g., national highway,                     The road network data are the basis for vehicle tracking and
state highway, city road, street etc.), name, speed information,             related applications. The road network model is represented
and the like. Digital map building is a process that utilizes the            with two-dimensional line segments.
information supplied by external equipments in terms of



                                                                        21                               http://sites.google.com/site/ijcsis/
                                                                                                         ISSN 1947-5500
                                                                          (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                                   Vol. 8, No. 3, 2010
   The road network model [5] is represented by the equation                           The Floating Car (Probe Car) technique is one of the key
                                                                                   technologies adopted by the ITS to get the traffic information
   Rn = (N, S),                                                                    in recent years [9]. Its basic principle is to periodically record
   N={n | n=(x, y), x, y Coordinates},                                             the location, direction, date, time and speed information of the
   S= {s | s = <m, n>, m, n N},                                                    traveling vehicle from a moving vehicle with the data of the
wherein Rn represents the road network, N represents the node                      Global Position System (GPS) as shown in Fig 2. The
set that indicates the coordinate point set of the road in the road                information can be processed by the related computing model
network which is a pair of longitude and latitude (x,y), S                         and algorithm so that the floating car data can be associated
represents the road segment set of the road network which is                       with the city road in real time [10]. This data can also be used
composed by the sequence <m, n>. S represents one directional                      as a source of data for creating research and commercial
road that has the beginning node m=begin(s) and the                                applications on vehicle tracking and road guidance systems.
termination node n=end(s). Fig. 1 shows a road network with
nodes, segments and an intersection point.                                                              III. RELATED WORK
                                                                                       The history of map making process starts with the
                                                                                   Egyptians who for the first time constructed a map for revenue
                                                                                   collection three thousand years ago. The digital map building is
                                                                                   a new concept developed after the revolution in Information
                                                                                   Technology. Though some work has been done in this area, a
                                                                                   number of map building techniques are being proposed to suit
                                                                                   the emerging requirements.
                                                                                       Y.L. Ip et al., have presented a technique for on-line
                                                                                   segment-based map building in an unknown indoor
                                                                                   environment from sonar sensor observations [4]. In their
                                                                                   approach, the environment data is first obtained from a rotating
            Figure 1. Basic elements of the road network model.
                                                                                   sonar sensor, analyzed and fed to the Enhanced Adaptive
    If there is a road segment sequence <si, sj,…, sk> in the                      Fuzzy Clustering (EAFC) module to extract the line segments
network Rn=(N, S), the termination node of every road segment                      within the workplace of robot. The basic motive is to use full
is the beginning node of the next road segment, this sequence is                   data set to obtain an initial approximation cluster centers via
called one directional road of the road network.                                   Fuzzy c-mean. This initial approximation helps reducing the
                                                                                   number of iterations required for Adaptive Fuzzy Clustering
                                                                                   (AFC). This approach is somewhat similar to Fast Fuzzy
B. Floating Car Data
                                                                                   Clustering (FFC) [11] which is a strategy to speed up the Fuzzy
    Nowadays, the main research focus in the community of                          c-mean (FCM). In order to facilitate the map building, the
Intelligent Transport Systems (ITS) is how to acquire real-time                    workplace of the robot is divided into squared areas as cells in
and dynamic transportation information. This information can                       order to extract the line segments. This mechanism reduces the
be applied in the transportation area like vehicle tracking,                       computation time when extracting the line segments within the
navigation, road guidance and so on. GPS is one of the system                      world frame. EAFC uses the Noise Clustering (NC) technique
which is used to provide real time information on moving                           proposed in [12] to extract the line segments. EAFC also uses
objects. It is a Satellite Navigation System which is funded and                   adaptive fuzzy clustering algorithm [13] and fast fuzzy
controlled by the U. S. Department of Defense [6, 7]. The                          clustering [11]. These algorithms are combined into a single
system consists of three segments viz., satellites that transmit                   algorithm with enhanced characteristics such as improvement
the position information, the ground stations that are used to                     in the computational burden and reduction of the effect of noisy
control the satellites and update the information, and finally                     data in fuzzy clustering algorithm. Besides, the authors have
there is the GPS receiver that computes its location anywhere                      proposed a compatible line segment merging technique to
in the world based on information it gets from the satellites [8].                 combine the similar line segments to a single long line segment
                                                                                   as a mechanism to reduce the number of segments in the world
                                                                                   model and further improve the quality of the map. This
                                                                                   technique is applicable for constructing maps of indoor
                                                                                   environment related to robotic applications.
                                                                                       Stefan Schroedl et al., have contributed to map-building by
                                                                                   introducing a system that generates digital road maps that are
                                                                                   significantly precise and contain descriptions of lane structure,
                                                                                   including the number of lanes and their locations, along with
                                                                                   detailed intersection structure[1]. The authors combine a large
                                                                                   quantity of possibly noisy data from GPS for a fleet of vehicles,
                                                                                   as opposed to a small number of highly accurate points
                                                                                   obtained from surveying methods. It is assumed that the input
    Figure 2. A moving vehicle (Floating car) fixed with a GPS receiver
                                                                                   probe data is obtained from vehicles that go about their usual




                                                                              22                               http://sites.google.com/site/ijcsis/
                                                                                                               ISSN 1947-5500
                                                                      (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                               Vol. 8, No. 3, 2010
business unrelated to the task of map construction, possibly                   receiver module was fixed in a moving car and the generated
generated for other applications based on positioning systems.                 NMEA sentences are stored in a log file in a laptop kept in the
The work of authors include the development of a spatial                       moving car. This GPS receiver generates $GPGGA, $GPGSA,
clustering algorithm for inferring the connectivity structure of               $GPRMC, $GPVTG and $GPGSV sentences at regular time
the map from scratch, the development of a lane clustering                     interval of one second. A list of NMEA sentences produced by
algorithms that can handle lane splits and merges and forming                  the GPS receiver and stored in a log file when travelled in a
an approach to inferring detailed intersection models. This                    road is given in Fig. 3.
system requires data from hundreds of vehicles already
connected to tracking systems for constructing a single segment                B. Refining the log file to get $GPRMC sentences
of the road.                                                                      The log file contains a number of different types of
    Thus a few number of digital map building techniques are                   sentences but the recommended minimum sentence C,
available but they suffer from high complexity of map building                 $GPRMC, provides the essential GPS PVT (Position, Velocity
process, requirement of more data and dependence on technical                  and Time) data. This data is used to locate moving objects in
skills of the person who is working with map building. The                     terms of latitude and longitude. The $GPRMC data format is
map building process proposed in this work addresses these                     given in Table I. The moving object, if attached with a GPS
problems. The process discussed here is a very simple one and                  receiver, can be located with the help of this NMEA sentence.
even a novice user without technical skills can easily create
route maps according to his requirements.                                                         TABLE I.        $GPRMC DATA FORMAT

                 IV.    MAP BUILDING PROCESS                                             Data Item           Format                   Description
                                                                                  Message ID             $GPRMC            RMC protocol header.
A. Collecting Floating Car GPS data in a log file                                 UTC Time
    GPS receiver communication is defined with National                           (Coordinated           hhmmss.sss        Fix time to 1ms accuracy.
                                                                                  Universal Time)
Marine Electronics Association (NMEA) specification. The
NMEA has developed a specification that defines the interface                                                              A Data Valid.
                                                                                  Status                 Char
                                                                                                                           V Data invalid.
between various pieces of marine electronic equipments. The
NMEA standard permits marine electronics to send information                      Latitude               Float             Degrees * 100 + minutes.
to computers and to other marine equipments [14] in                               N/S Indicator          Char              N=north or S=south.
predefined formats. Most computer programs that provide real
time position information recognize data that are in NMEA                         Longitude              Float             Degrees * 100 + minutes.
format which includes the complete latitude, longitude,                           E/W Indicator          Char              E=East or W=West.
velocity and time computed by the GPS receiver. In NMEA                           Speed over Ground      Float             Speed Over Ground in knots
specification system, the collected GPS data is converted into a
                                                                                                                           Course Over       Ground     in
line of text, called a sentence, which is totally self contained                  Course over Ground     Float
                                                                                                                           Degrees
and independent from other sentences. The commas act as
                                                                                  Date                   ddmmyy            Current Date
terminators for the sentences and the programs that read the
data should only use the commas to determine the end of a data                    Magnetic Variation     Blank             Not Used
item.
                                                                                  E/W Indicator          Blank             Not Used
                                                                                  Mode                   Char              A Autonomous
                                                                                  Checksum               *xx               2 Digits
                                                                                  Message
                                                                                                         <CR><LF>          ASCII 13, ASCII 10
                                                                                  Terminator


                                                                                  An example of $GPRMC NMEA sentence is given below:


                                                                                  $GPRMC,120642.206,A,1118.4253,N,07742.4325,E,31.6,
                                                                               317.52,140510,,,A*62
                                                                                  Where
                                                                                  $GPRMC       : Recommended Minimum sentence C
                                                                                  120642.206 : Fix taken at 12:06:42.206 UTC
   Figure 3. Log file of floating car GPS data with $GPGGA, $GPGSA,               A            : Status A=active or V=Void.
               $GPRMC, $GPVTG and $GPGSV sentences                                1118.4253,N : Latitude 11 deg 18.4253' N
                                                                                  07742.4325,E : Longitude 77 deg 42.4325' E
    In order to collect the floating car GPS data, Wonde-X                        31.6         : Speed over the ground in knots
series GPS receiver (ZX4125) module was used. This GPS                            317.52       : Course over the ground



                                                                          23                                     http://sites.google.com/site/ijcsis/
                                                                                                                 ISSN 1947-5500
                                                                          (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                                   Vol. 8, No. 3, 2010
    140510            : Date – 14th of May 2010                                            After obtaining the points at road centre lines after
    A                 : Autonomous mode                                               interleaving m sentences in the middle, the slope between two
    *62               : The checksum data, always begins with *                       adjacent points say xi+1,yi+1 and xi,yi is calculated. Now an
                                                                                      imaginary line perpendicular to the slope is drawn through xi,yi
                                                                                      and they are made to intersect an imaginary circle of radius 'r'
                                                                                      on both sides as shown in Fig. 5. The points of intersection on
                                                                                      either side (lxi, lyi and rxi, ryi) are recorded and they act as the
                                                                                      nodes for the left line of the road and right line of the road,
                                                                                      forming road segments. This process is repeated for all values
                                                                                      of 'i' ranging from 1 to k-1. It is to be noted that the radius of
                                                                                      the imaginary circle determines the width of the road segment.
                                                                                      Roads with different widths can be drawn by altering the radius
                                                                                      'r' of the imaginary circle.
                                                                                          The following algorithm is used to extract segments on
                                                                                      either side of the road centre line. This algorithm extracts
                                                                                      segments by computing road vectors on either side of the road
                                                                                      and adds lines in vectors. The output of this algorithm for a
                                                                                      sample data is given in Fig 5.
              Figure 4. Refined Log file of $GPRMC sentences                             Input : Refined log file; sentence count m; radius of
                                                                                      imaginary circle r; number of sentences to be interleaved n.
    Hence the next step in map making process is to refine the
                                                                                          Output : Line segments on the left and right side of the road
log file by removing other sentences in such a way that it
                                                                                      forming a road segment; line vector lx, ly for road left line and
contains the $GPRMC sentences only as shown in Fig 4. This
                                                                                      rx, ry for road right line.
refined log file now contains the path of the probe car in terms
of latitude and longitude at an interval of one second per                               Step 1 : Initialize data items: m=0; i=0; n=40, r=0.5.
sentence.                                                                                Step 2: Repeat step 3 to step 4 till the end of log file.
                                                                                         Step 3: Read the next sentence from the log file
C. Segment Extraction                                                                    Step 4: If m > =n
    The road segment extraction is done by considering the                                          Read the sentence from the log file and
locations of the probe car at a fixed time interval. Since in most                                  store longitude into xi and latitude into yi.
cases there is no significant distance between the probe car and                                    i=i+1
the road centre line, it is assumed that the path of the vehicle is                                 m=1
along the road centre line. The idea is to pick out 'k' locations in                           Else
road centre line continuously by interleaving 'n' $GPRMC                                            m = m +1
sentences. When this process is iterated, the ith location of the                        Step 5 : k = i-1
moving vehicle xi,yi (longitude xi and latitude yi) for all the                          Step 6 : Repeat step 7 for i =1,2,….,k-1
values of 'i' ranging from 1 to 'k' along the road centre line at                        Step 7 :
'm' seconds interval is obtained.                                                                dx=xi+1 - xi
                                                                                                 dy=yi+1 - yi
                                                                                                 slope_radians =tan-1(dy/dx)
                                                                                                   i =slope_radians / ( /180)
                                                                                                   = i + 90
                                                                                                 left_plot_radians = ( /180) *
                                                                                                 lxi = xi + r*cos(left_plot_radians)
                                                                                                 lyi = yi + r*sin(left_plot_radians)
                                                                                                 Plot(lxi, lyi)
                                                                                                   = + 180
                                                                                                 right_plot_radians = ( /180) *
                                                                                                 rxi = xi + r*cos(right_plot_radians)
                                                                                                 ryi = yi + r*sin(right_plot_radians)
                                                                                                 Plot(rxi, ryi)
                                                                                         Step 8 : line(lx,ly) - Add the line in vectors lx and ly to the
                                                                                      current axes.
                                                                                         Step 9 : line(rx,ry) - Add the line in vectors rx and ry to the
                                                                                      current axes.
                                                                                         Step 10 : Stop.
Figure 5. Obtaining Points for Left and Right Road lines using a Point of Road
                                  Centre line




                                                                                 24                                http://sites.google.com/site/ijcsis/
                                                                                                                   ISSN 1947-5500
                                                                           (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                                    Vol. 8, No. 3, 2010
                                                                                      process is iterated for subsequent segments until all the
                                                                                      segments are processed.
                                                                                          The algorithm used for segment merging is given below.
                                                                                          Input : Negligible slope (angle) , road segments (vectors
                                                                                      for left line lx,ly and right line rx,ry).
                                                                                          Step 1 : Compute the slope(angle) of the line connecting
                                                                                      xi,yi and xi+1,yi+1 of two adjacent segments ( i and i+1) with
                                                                                      respect to 'x' axis.
                                                                                          Step 2 : Repeat step 3 till all the segments are processed
                                                                                          ( i=1,2,…,k).
                                                                                          Step 3 : if abs( i i+1) < , merge the two segments i.e.,
                                                                                      remove intermediate node; otherwise do not merge.




 Figure 6. Road segments extracted on the left and right sides of road centre
                         line (vehicle trajectory)


D. Segment Merging
    The road segment merging technique is used to merge the
similar basic road segments together to form a single road
segment. It is observed from the outcome of the segment
extraction algorithm that a number of adjacent road segments
are similar in direction and they can be merged together to
form a single segment. The primary advantage of segment
merging is that it reduces the number of nodes in the road
network and hence simplifies the map. It also results in faster
generation of the map because of the less number of segments
in the map.

                                                                                      Figure 8. Road segments with left and right road lines after removal of vehicle
                                                                                                                        trajectory




 Figure 7. Road segments extracted on the left and right sides of floating car
                     GPS data after segment merging

    The criteria used to merge the road segments is the slope                                    Figure 9. Road Map after removing the plots for nodes
(angle) i of the line connecting xi,yi and xi+1,yi+1 with respect
to x axis. A threshold limit is set for the negligible slope                              Thus the segment merging algorithm is based on negligible
(angle) and if this slope(angle) is within the threshold, say 5°                      slope and it merges the similar segments together which results
for two adjacent segments, then both the segments are merged                          in faster production of digital maps by eliminating unwanted
together, otherwise the segments are left as they were. This                          nodes in the road network. The extracted segments are shown




                                                                                 25                                    http://sites.google.com/site/ijcsis/
                                                                                                                       ISSN 1947-5500
                                                                           (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                                    Vol. 8, No. 3, 2010
in Fig. 6 and the merged segments are shown in Fig 7. It is                            •    Removal of a segment: Sometimes, a segment at the
observed that different threshold values for negligible slope                               end of one road and a segment at the beginning of the
(angle) can be used to get required accuracy or width of the                                next road may intersect and cross one another. In that
route map. The merged segments after removing vehicle                                       case, one of the segments is deleted.
trajectory are shown in Fig. 8. The final map of the road after
removing the plots for nodes is shown in Fig. 9.                                       •    Extending a segment: When two or more segments do
                                                                                            not join at the intersection, the segments are extended
                                                                                            based on the previous slope till they form refined
E. Refining Intersection of Road Segments
                                                                                            intersection. It is to be noted that extending a segment
   The obtained road network after segment merging is still                                 is different from adding a segment.
unrefined at intersections of road segments. The refinement of
road intersection is done in the following steps.                                       A map with unrefined road segments intersection is shown
                                                                                    in Fig. 10. The road intersection is refined and Fig. 11 shows
                                                                                    map with refined intersection of road segments. This process is
                                                                                    repeated for entire set of road intersections till a fine-tuned map
                                                                                    is obtained.

                                                                                    F. Labeling Points of Interest
                                                                                        A map contains points of interest which means places
                                                                                    which are important and noted down on the maps. Placing text
                                                                                    on a map is a particularly difficult challenge in digital maps
                                                                                    because the axis of the digital maps can be changed
                                                                                    dynamically. This is the final step in this map making process
                                                                                    and points of interests are noted down from the probe car data
                                                                                    and the text is placed so that it is readable and easily located.
                                                                                    Care has also been taken that the text does not interfere with
                                                                                    the map’s data or design. Different font types, styles, sizes, and
                                                                                    colors are used to establish clear association between text and
                                                                                    map features like telephone post, petrol bunks and toll gates.
                                                                                    Fig. 12 displays a legible point of interest.

    Figure 10. Digital Map with unrefined intersection of road segments




                                                                                              Figure 12. Digital Map with Labeled Points of Interest

                                                                                                    V. EXPERIMENTAL RESULTS
       Figure 11. Digital Map with refined intersection of road segments                The outcome of the map making process is a list of
                                                                                    segments drawn on the map with refined intersections of roads.
   •      Adding a segment: Many a times, there may be a                            For the purpose of demonstration, the floating car data was
          necessity to add one segment either at the end or at the                  collected in different roads. Segments for the roads are
          beginning of a road so that the road is connected to                      extracted using the segment extraction algorithm. Thereafter
          another road. In that case, a new segment is added.                       segment merging was done based on the slope of the adjacent
          Sometimes, instead of adding an entire segment, only                      road segments. Finally, the road intersection points are refined
          the left or right line of the road segment is extended.                   and points of interest are labeled to get the final map. The




                                                                               26                                  http://sites.google.com/site/ijcsis/
                                                                                                                   ISSN 1947-5500
                                                                        (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                                 Vol. 8, No. 3, 2010
following experiments demonstrate the simplicity and accuracy                      more accurate segments depending on the necessity. Two
of the map making.                                                                 different segment extractions for the same floating car data
                                                                                   with different values of 'r' according to varying requirement is
A. Simplicity                                                                      shown in Fig. 15, 16. It is to be noted that the nodes are shown
    A comparison of the map generated by the proposed map                          only to differentiate segments and they are removed in the final
making process as shown in Fig. 13 versus a digital map                            map. A more accurate and thinner road is shown in Fig. 16
available in the web as shown in Fig. 14 indicates that the                        compared to the one shown in Fig. 15.
proposed process is simple and the map includes only the
desired routes and points of interest. The generated map can be
easily interpreted because of its simplicity.




                                                                                              Figure 15. Map with thick roads with less accuracy




 Figure 13. Map generated by the digital map building process discussed in
                                this paper




                                                                                      Figure 16. Map with thin roads for the same data with more accuracy


                                                                                       Accuracy can also be obtained by making the sampling
 Figure 14. A previously available digital Map with dense routes and places        sentences interval as minimum. Table II shows a Comparison
                                                                                   of segments extracted and segments generated after merging at
B. Accuracy                                                                        different intervals of sampling sentences. Accuracy increases
    One of the input parameter of segment extraction algorithm                     with more number of extracted segments and merged segments.
is the radius of the imaginary circle 'r'. By adjusting the value                  The graph shown in Fig. 17 gives a comparison between the
of 'r' the road segment may be made thicker or thinner to get



                                                                              27                                  http://sites.google.com/site/ijcsis/
                                                                                                                  ISSN 1947-5500
                                                                                                                          (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                                                                                   Vol. 8, No. 3, 2010
number of extracted and merged segments for a floating car                                                                            This process can be used to create maps but the process
data at different sampling intervals.                                                                                              depends on a probe car for data collection. In future, this
                                                                                                                                   drawback can be eliminated by collecting GPS data from the
                                                                                                                                   GPS enabled vehicles already connected with commercial
                                  TABLE II.    COMPARISON OF SEGMENTS EXTRACTED AND SEGMENTS
                                   GENERATED AFTER MERGING AT DIFFERENT INTERVALS OF SAMPLING                                      applications. This work can also be extended to handle roads
                                                          SENTENCES.                                                               with multiple lanes.

 Interval                                        No of
                                                                                     No. of             Percentage of                                              REFERENCES
    of                                           Points                 No. of
                                                                                   segments             reduction in               [1]    Stefen Schroedl, Kiri Wagstaff, Seth Rogers, Pat Langley and
sampling                                       generated              segments
                                                                                     after             segments after                     Christopher Wilson., “Mining GPS traces for Map Refinement,” Data
sentences                                      by probe               extracted
                                                                                   merging                merging                         Mining and Knowledge Discovery, vol. 9, pp. 59-87, 2004.
(in No. s)                                      car data
    20                                            413                    19               9                   52.63                [2]    D. Lee, “The Map-Building and Exploration Strategies of Simple
    40                                            413                    8                4                   50.00                       Sonar-Equipped Mobile Robot,” Cambridge Univ. Press., Cambridge,
    60                                            413                    4                2                   50.00                       1996.
    80                                            413                    3                2                   33.33                [3]    P. Weckesser, and R. Dillmann, “Modeling unknown environment with
                                                                                                                                          a mobile robot,”, Robotics Autonomous Systems, vol. 23, pp. 293–300,
                                                                                                                                          1998.
                                                                                                                                   [4]    Y. L. Ip, A.B. Rad, K.M. Chow and Y.K. Wong, “Segment-based map
                                  30                                                                                                      building using enhanced adaptive fuzzy clustering algorithm for mobile
                                                                                                                                          robot applications,” Journal of intelligent and robotic systems, vol. 35,
                                                                                                                                          pp. 221-245, 2002.
Num ber of Segmen ts G enerated




                                                                                                                                   [5]    LÜ WeiFeng1, ZHU TongYu, WU DongDong, DAI Hong and HUANG
                                                                                                                                          Jian, “A heuristic path-estimating algorithm for large-scale real-time
                                  20                                                                                                      traffic information calculating,” Science in China Series E:
                                                                                                                                          Technological Sciences, vol. 51, pp. 165-174, Apr. 2008.
                                                                                                                                   [6]     B.W. Parkinson, and J.J. Spilker, “Global Positioning System: Theory
                                                                                                                                          and Applications,” American Institute of Aeronautics and Astronautics,
                                  10                                                                                                      Washington, 1996.
                                                                                                                                   [7]    Interface control document. Navstar GPS Space Segment (Navigation
                                                                                                                                          User Interfaces), 2000.
                                                                                                                                   [8]    Asoke K Talukder and Roopa R Yavagal. “Mobile Computing-
                                                                                                                                          Technology, Applications and Service Creation,” Tata McGraw Hill
                                   0
                                                                                                                                          Publishing Company, 2005.
                                              20                  40                 60                    80
                                                                                                                                   [9]    X. W. Dai, M.A. Ferman, “A simulation evaluation of a real-time traffic
                                                             Sam pling Sentences Interleave d
                                                                                                                                          information system using probe vehicles,” IEEE ITSC, vol. 1, pp. 12-15,
                                                                                                                                          2003.
                                          No. of Extracted Segments                       No. of Segments after Merging            [10]   R. Kuehne, R.P. Schaefer and J. Mikat, “New approaches for traffic
                                                                                                                                          management in metropolitan areas,” IFAC CTS, 2003.
                                  Figure 17. Number of segments generated at different sampling intervals
                                                                                                                                   [11]   T.W. Cheng, D.B. Goldgof, and L.O. Hall, “Fast fuzzy clustering,”
                                                                                                                                          Fuzzy Sets Systems, vol. 93, No. pp. 49–56, 1998.
    The accuracy of the map is inversely proportional to the                                                                       [12]   R.N. Dave, “Characterization and detection of noise in clustering,”
                                                                                                                                          Pattern Recognition Lett. vol. 12, pp. 657–664, 1991.
radius of the imaginary circle which is used to draw the road,
negligible slope and interval of sampling sentences. Desired                                                                       [13]   R.N. Dave, “Use of adaptive fuzzy clustering algorithm to detect lines in
                                                                                                                                          digital images,” Intelligent Robots Computer Vision, vol. VIII, pp. 600–
accuracy can be obtained by adjusting these values according                                                                              611, 1989.
to the need.                                                                                                                       [14]   National Marine Electronic Association,            http://www.nmea.org,
                                                                                                                                          accessed on 20.04.2010.
             VI. CONCLUSION AND FUTURE WORK
                                                                                                                                                                AUTHORS PROFILE
    This paper introduces a customized digital road map
building process which can be used to build digital maps of
desired routes. By changing the input parameters, the accuracy                                                                     G. Rajendran is a second-year Doctoral
                                                                                                                                   Research Scholar in the Research and
of the route can be altered to required level and route maps for                                                                   Development Centre of Bharathiar University.
different types of roads, for instance, national highways, state                                                                   He received his Masters degree in Computer
highways, city roads and streets can be drawn. The results of                                                                      Applications and M.Phil degree in Computer
the map building process discussed in this paper are compared                                                                      Science from Bharathiar University. He passed
with other maps generated by other commercial making                                                                               the National Eligibility Test for lectureship
                                                                                                                                   conducted by UGC, the apex body of higher
software and it is found that the proposed process is simple, yet                                                                  education in India. He is also working as an
powerful. This map making process has eliminated the                                                                               Assistant Professor of Computer Science at
complexity of the previous works carried out in this area and                                                                      Thiruvalluvar Government Arts College,
customized road map can be built for commercial and other                                                                          Rasipuram, India. His research interests include
applications like vehicle tracking, navigation and route                                                                           Mobile Computing, Data Mining and programming-language support for
guidance systems.                                                                                                                  massive-scale distributed systems.




                                                                                                                              28                                      http://sites.google.com/site/ijcsis/
                                                                                                                                                                      ISSN 1947-5500
                                                                       (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                                Vol. 8, No. 3, 2010
Dr. M. Arthanari holds a Ph.D. in Mathematics
from Madras University as well as Masters
Degree in Computer Science from BITS,
Pilani. He was the professor and Head of
Computer Science and IT Department at Tejaa
Shakthi Institute of Technology for Women,
Coimbatore, India. At present he is the
Director, Bharathidhasan School of Computer
Applications, Ellispettai, Erode, Tamilnadu. He
holds a patent issued by the Govt. of India for
his invention in the field of Computer Science.
He has directed teams of Ph.D. researchers and
industry experts for developing patentable products. He teaches strategy,
project management, creative problem solving, innovation and integrated new
product development for last 36 years.

M. Sivakumar has 10+ years of experience in
the software industry including Oracle
Corporation. He received his Bachelor degree
in Physics and Masters in Computer
Applications from the Bharathiar University,
India. He is currently doing his doctoral
research in Anna University, Coimbatore. He
holds a patent for his invention in embedded
technology. He is technically certified by
various professional bodies like ITIL, IBM
Rational Clearcase Administrator, OCP -
Oracle Certified Professional 10G, PRINCE2 and ISTQB. His research
interests include Embedded Technology, Ubiquitous Computing and Mobile
Computing.




                                                                              29                            http://sites.google.com/site/ijcsis/
                                                                                                            ISSN 1947-5500