Realtime, Personalized Path Prediction and Automobile Accident Detection System

1
Real-time, Personalized Path Prediction and
Automobile Accident Detection System

• Tiffany Burrell
• Southern University and A M College
• Baton Rouge, LA
• Center for Urban Transportation Research
• University of South Florida

2
TRACIT Project

• TRACITs objective is to develop a more effective
  way to collect data about peoples travel
  behavior in order to help the traffic engineers
  and civil engineers determine what would be the
  most efficient road plan for a specific area.
• Software has been developed to put on GPS-enabled
  mobile devices to collect information about a
  persons travel behavior.
• Currently the project is at the point of
  improving the automation of the software package.

3
Global Positioning System

• GPS System
• 27 Satellites
• Orbit the Earth twice a day
• Equipped with an Atomic clock
• GPS Receiver
• 3-D Trilateration
• GPS Receiver locates 4 satellites above it
• Then it calculates the distance to GPS satellites
  by timing a signal's journey from satellite to
  receiver

4
My Research Assignment

• THE PROJECT
• Develop code that will predict a client's path at
  runtime without user input.
• Warn the client of up coming traffic that they
  may encounter.
• THE PURPOSE
• To reduce traffic congestion caused by accidents
• Save time
• Reduce the number of irrelevant traffic warnings
  received from a traffic detection subscription
• Reduce the number of times people say, Had I
  known there was going to be traffic, I would have
  gone the other way.

5
Requirements

• In order to predict someones travel path, their
  travel behavior should be recorded for at least
  two weeks using GPS technology.

6
Collection of Coordinates
7
Create a Polyline

• There is a polyline created for each TripID

8
Place a Buffer Around the Polyline

• The Buffer is like a fence around a house. It is
  there to detect when someone is within your
  property.

9
Methodology

• Every five points the method is invoked
• Query all polygons of the client that the short
  path lies within and that occurred within a hour
  of the current time
• Detect all accidents that lie within those
  polygons
• Send the client a text message that describes the
  description of the location of the accident

10
Methodology
11
Methodology Continued
12
Methodology Continued
13
Static Testing

• Purpose
• Make sure the buffers were large enough
• Detect active automobile accidents
• Make sure the correct polygons were queried
• Detect other logical errors in the code
• Test the time and user id factor of the algorithm
• Material
• Java Thread
• 20 Accidents
• 2015 Trips
• 15 short paths
• Results
• Passed

14
Dynamic Test

• Purpose
• Determine if the user would receive the warning
  in ample time
• Test spatial reference
• Material
• GPS enabled cellular phone
• TRACIT Software
• Java thread
• Results
• Due to a problem between the ArcObject and Java
  interface the code could not detect users travel
  path at runtime. Parts of the application will
  have to be redone using ArcServer.

15
Conclusion

• It is our hope that once people receive
• the accident warning, they would
• choose an alternate route in order to
• decrease traffic congestion within the
• vicinity of the accident.

16
Future Endeavor

• When an accident is detected along a clients
  most frequently pursued route, TRACIT will
  provide them with an alternate route.

17
Acknowledgements

• University of South Florida
• Dr. Miguel Labrador
• Mentor Sean Barbeau
• Narin Persad-Maharaj
• Alfredo Perez
• Center for Urban Transportation Research
• National Science Foundation
• Science Engineering Alliance (SEA)
• Timbuktu Academy/ LS- LAMP
• Ronald E. McNair Program

18
Questions?For more information
about TRACIT please contact Sean Barbeau at
barbeau_at_cutr.usf.edu