1. Simulating Traffic Congestion on Route 1 Jimmy O'Hara Computer Systems Lab 2009-2010

2. Abstract
During rush hours traffic on Route 1 is at an almost standstill, the road is not capable of holding the amount of traffic that travels on it during peak hours
Purpose is to first create a realistic simulation of this congestion using a multi-agent system, with cars interacting correctly with each other and the road system
Part two of the project is to use VDOT data to find the cost per driver of the current system
Part three of the project is to use this output as a control then modify the system in order to find a viable solution to the congestion problem

3. Introduction
Through the use of accurate traffic count data from the Virginia Department of Transportation, the project will create a realistic simulation of the current traffic situation on Route 1 to scale with the Java programming language
The simulation is created through the use of a multi-agent system. This allows for the cars to interact with each other while still following their own separate goals
The program will then analyze the traffic count data in order to determine the cost per driver through the following factors: average speed, cost of gas, and travel time
The simulation will then be modified through solutions such as: naked intersections, round-a-bouts, additional lanes, etc. that are supposed to increase the flow of traffic in this type of traffic system, trying to find a viable solution to the traffic problem

4. Background
This type of project has been completed in Turkey by Atilla Elci and Ali Zambakoglu who simulated a city traffic system in order to determine how to increase the flow of traffic in the system
Previous research in this at TJHSST in has been completed by Craig Haseler, Timmy Galvin, and Paul Wood. Most of these projects were written with the Java programming language
Similar research into this field was completed by Xu Jin, Mhamed Itmi, and Habib Abdulrab who used a multi-agent system to create a smart bus system, which reduced the number of buses driving in the city traffic system, thus improving traffic flow
Similar research into this field was completed by Pierrick Tranouez, Patrice Langlois, and Eric Duade who created a multi-agent urban traffic simulation in order to allow the system to learn how to deal with everyday traffic congestion. They attempted to minimize the cost per driver using local knowledge of the road system.

5. Fig. 1: Figures, graphs for analysis
Stretch of road
Avg. Daily Traffic
% Cars
% Buses
% Rush Hour
% Peak Direction
FXC Parkway to Woodlawn Rd
42000
98% 2%
0.0825
0.6865
Woodlawn Rd to MV Memorial HW
39000
0.0771
0.5539
MV Memorial HW to MV HW
35000
0.0790
0.5257
MV HW to Kings HW
60000
0.0743
0.6560
Kings HW to SCL Alexandria
45000
0.0656
0.6879
Fig. 1: Figures, graphs for analysis

6. Discussion
Part 1:
Graphics part, it will be the easiest part for me to finish and will be a good way to re-introduce myself to Java
This includes creating the road system to an accurate scale,creating the cars, and having them behave as actual cars.
The multi-agent system is placed inside a grid matrix that allows the cars to “see” their own surroundings and make their own traffic decisions in order to navigate to their final destination.
This part will be tested mostly through an eye test, in other words is the car acting like a normal car would, does the car move to the correct lane to turn off the road and does it turn off the road realistically.

7. Discussion
The car driving on Route 1, the system is simplified to being only one way

8. Discussion Part 2 This is the data part of the project
Will incorporate reading in the traffic count data from the VDOT data I found and using the data to create an accurate visual representation of what is happening on the road
Then producing an output of the cost per driver in using the road, which will be analyzed through a variety of factors such as: average speed, cost of gas, and travel time
This part can be tested by comparing the visual simulation to the reality on Route 1, as well as comparing the cost output to what the visual representation seems to say

9. Discussion
Part 3
This is where the system will be modified to include the various ways the road can be improved
Solutions include: naked intersections, round-a-bouts, additional lanes, and designating lanes for public transportation only
These will be based on user options, allowing the user to create the system they want. This will ensure that the each method can be tested separately as well as together with
other solutions, for example creating an extra lane and making intersections and round-a-bouts in the same simulation.
This section cannot be completely tested, because it is the more theoretical aspect of the project. Although the code can be tested to make sure the cars react to the different systems
correctly, there is no real life situation with which to compare the results.

10. Results
Currently the matrix has been created and the multi-agent system is working for the most part.
Currently the cars drive down a one-way, multi-lane road, following their own route in order to reach their destination.
The 2.0 version of this is having the system reverse on itself, so that the simulation is an accurate portrayal of the two-way Route 1.
The system successfully has the cars driving both directions, and interacting with the other cars on their side of the road. The carsare not completely able to turn left.

11. Expected Results and Benefits
I expect to determine the best course of action to alleviate the traffic through my program. With this result the project will be of great value not only to the people working on a solution to Route 1’s congestion, but also to similar road situations around the world that are experiencing the same problems as Route 1.
Also the residents of Alexandria who use Route 1, like myself, will find great value in this project as it informs them on how changing the way Route 1 flows would affect their travel times and other factors.