Transcript see Narin`s poster - Computer Science and Engineering

REU 2007-Real Time Path Prediction
Narin Persad, Mentor: Sean Barbeau
Introduction
Results
Due to temporary technical complications with ESRI’s GIS software at
the time of testing , synthetic trips were used to test the algorithm.
Various simulations using the synthetic trips were performed to
anticipate what might happen in real-time situations.
Advances in GPS navigation technology has given motorists
exceptional control over their route and travel
information. These devices, however, require effort on
behalf of the user to input their destination and they have
little or no benefit beyond the confines of an automobile.
Additionally, the high cost of these devices is often too
much of an expense for the average consumer.
The motivation of this research is to give a commuter an
affordable and practical alternative to expensive GPS
navigation devices.
The objective of this research is to develop an algorithm
that uses a commuter’s travel patterns to automatically
determine their destination. Messages concerning route
hazards and alternate routes can then be sent via a GPSenabled cell phone.
Methodology
In order to record travel patterns, GPS-enabled cell
phones were distributed among volunteers
These cell-phones transmit coordinates, duration
and time of travel as well as other information
•A SQL database stores and organizes this information
by user and trip ID.
•Use ESRI’s ArcObjects Java API to build a polyline,
connected line segments, from the coordinates in the
SQL database.
Buffer the polyline to create a polygon that has
area and store this shape in ArcSde, a GIS server.
•Catalogue the user’s path history in ArcSde and any
accident points
Situation 1
Fig 1.
polylines
The first two coordinates of
the polyline did not lie within
the buffer. The algorithm
deleted the first two
coordinates until the polyline
was enclosed within a buffer.
Fig 2. A polygon layer as stored in
ESRI’s ArcSde GIS database
Algorithm Flowchart
Situation 2
An offset coordinate is
created from the last
coordinate and buffered.
The algorithm found a trip
buffer.
Conclusion
Despite the technical complications and lack of real-time testing,
the algorithm behaved exactly as expected with the synthetic trips.
The results of the simulations show promise for future real-time use.
Future Work
To further improve the algorithm , plans to integrate other trip
properties are being developed. Specifically, to narrow the destination
possibilities, time of departure and arrival is being incorporated into
the algorithm to account for the traveler’s schedule.
References
•ESRI documentation and support: http://www.esri.com/
•Kyriakakos, M. Hadjiefthymiades, S. Frangiadakis, N. Merakos, L.
Dept. of Informatics & Telecommun., Athens Univ., Greece, Multi-user
driven path prediction algorithm for mobile computing
Department of Computer Science & Engineering