Here is a pdf file of my paper Capstone Paper – Emily R Smith

below is an un-formatted version of the paper for the blog

Microscopic Traffic Simulation of Dublin Ohio 

created with Aimsun Next

Emily R Smith

The Ohio State University

Abstract

My project focused on data visualization of a single intersection. My original plan was to create this model from scratch using  Java a general-purpose programming language, however this plan failed as I lacked the knowledge and the time to create a program so different from my previous work. From this setback I decided to create the intersection using industry quality software. This software turned out to be Aimsun which is used by government agencies, and academic institutions, this includes The New York Department of Transportation and MIT, Massachusetts Institute of Technology. Because, I didn’t create the software I decided to do some data collection in the field on the intersection that I modeled. I collected data on the amount of cars, and what direction they moved through the intersection, I also timed the traffic light states. Finally I created the intersection in the Aimsun software using all the data I collected from the intersection plus the speed and frequency data from the City of Dublin to check that my vehicle per hour measurement was realistic and to give realistic speeds to the cars. I adjusted the simulation until it was a good model of that intersection during the hour and a half time period that I collected data on site. By completing this model of a high traffic single intersection I learned so much about the process of creating realistic models of traffic flow and how complex traffic management is.

Microscopic Traffic Simulation of Dublin Ohio 

created with Aimsun Next

Traffic is something that I am very familiar with being a commuter student , however I wanted to learn more about traffic and wanted to visualization the motion of individual vehicles. I decided to achieve this my modeling a microscopic part of Dublin Ohio’s traffic system. The intersection I choose was high traffic and was complex with multiple lanes and separate left turn lanes. My original plan for the final product was to create software to model traffic during a specific time frame. My final project however uses traffic modeling software created by Aimsun, instead of a software of my own creation.

Literature review

In reviewing the Ohio Public Safety website, you quickly get a feel for how important traffic flow is to the lives of citizens. In reviewing traffic crash information, including the number of fatalities each year, it is clearly evident that proper traffic flow can literally mean the difference between life and death. This alone is a significant validation for the importance of traffic analysis.

In addition, I reviewed several academic papers on the study of traffic flow and the use of software simulations. The traffic flow simulation created by Martin Treiber was the starting point of much of this research, it is a great example of how software can be used to recreate life like situations. From there I found a book Treibar has co written called Traffic Flow Dynamics that discusses vehicle based flow dynamics and how simulation can be used at the microscopic and macroscopic levels. This book makes traffic an interdisciplinary area of interest, as it deals with human behavior, computer science, physics, and traffic engineering. Therefore my personal research understanding how these fields can combine is important because it will allow me to join the conversation on these complex interdisciplinary topics with greater understanding. 

Another work also co written by Martin Treiber is the article “General Lane-Changing Model MOBIL for Car-Following Models” really tied research of traffic flow to the safety of the roads, stating that “drivers’ lane-changing behavior has a direct influence on traffic safety” Kesting, A., Treiber, M., & Helbing, D., 2007, p. 1). This just furthers how important traffic research.

Methodology

After extensively researching the traffic flow analysis I wished to do, I realized that, due to time constraints, and the existence of adequate analysis applications, it would be more effective for me to review and choose an existing software tool, rather than attempting to recreate one. In order to complete the microscopic traffic model I first had to choose a site from which I would collect traffic data. I chose a high traffic intersection to provide a larger sample of data. I then needed to collect the data and transfer it to my spreadsheet (see Appendix A and Appendix B for the spreadsheets) , so that I could calculate variables, including vehicles per hour, percentages of turners, and average traffic light time. I then inputted the data into Aimsun Next to create a representation of the intersection.

To create the model, I acquired a reference photo of the selected intersection. I used the image to add the lanes surrounding the intersection, making sure to add left turn lanes. I then had to create nodes to represent the directions of travel through the intersection. I inputted the number of vehicles that traversed each lane. I then added the traffic controller, the device that regulates the flow of traffic through the intersection. Following that, I imported the speed data from the City of Dublin Division of Police for this particular intersection (Traffic/Speed Survey Results by StealthStats, 2018). The data was from March 27th to 29th of 2018 on Bridge Street. With all of the setup work done, I had to choose the simulation scenario I wanted. I then ran the simulation through the software. 

Figure 1. An image of the simulation in progress. Taken from a screenshot of the Aimsun Next simulation software.

Results

The resulting traffic model was an adequate representation of the data, although it did not fully represent the actual traffic that flowed over time. This was primarily due to the manual collection method I was able to use. This method provided more limited data. Had I been able to use actual traffic flow meters, I would have had a more realistic representation because of the increased accuracy and quantity of data that could have been collected. Overall, the model was successful in helping me understand the process of traffic analysis.

Conclusion

 I believe my results from my microscopic traffic study gave me a much clearer insight into how traffic data may be collected, modeled, and simulated. I also have a clearer understanding of how this information can be used to improve the systems in place to control the overall flow of vehicles through a populated area. I am confident that this experience has helped me develop more skills related to problem solving and original inquiry.

References

Kesting, A., Treiber, M., & Helbing, D. (2007). General Lane-Changing Model MOBIL for Car-Following Models. Transportation Research Record: Journal of the Transportation Research Board,1999(1), 86-94. doi:10.3141/1999-10

Ohio Department of Public Safety. (n.d.). Retrieved from https://publicsafety.ohio.gov/wps/portal/gov/odps/home

Traffic/Speed Survey Results by StealthStats. (2018). Retrieved from https://dublinohiousa.gov/dublin-police/trafficspeed-survey-results-by-stealthstats/

Treiber, M. (n.d.). Microsimulation of Traffic Flow: Onramp. Retrieved from http://www.traffic-simulation.de/

Treiber, M., & Kesting, A. (2013). Traffic flow dynamics: Data, models and simulation. Heidelberg: Springer.

Appendix A

Frequency of Vehicles during one Traffic Cycle from 6/28/2019

Appendix B

Traffic Lights Time Averages from 6/28/2019

Conclusion:

I finished the intersection in Aimsun Next. I used the data I collected mostly, but I also used Google Maps for a reference of the intersection, and the data collected by the city to check that my data wasn’t too skewed, as well as adding speed data. Aimsun is a software that is used in the industry and I was able to get a free version for educational purposes. This software was comfortable to use and had almost all the features I needed for a completely realistic intersection. The only thing that it didn’t have was control to change the length of the yellow lights for each traffic light. I had to use a fixed length for the whole intersection. I also wasn’t able to add pedestrian crossings to this model. That feature is only made available for the full version of the software. Even with these minor setbacks, the intersection is a good representation of the real intersection and was easy software to use. Below is an image of the intersection during the simulation with, and without, the Google Maps reference image.

Overall, I learned so much from this project, like how hard it is to collect data on cars by hand, that traffic systems are much more complex than I first realized, and that it is sometimes necessary to change plans if needed — the whole purpose of the project was to the learn about something that is not my major but still interested me.

The experience also reminded me of my internship with the city of Dublin Engineering Department during my senior year of High School. There I got to work on GIS road creation, data entry, and even got to talk with the man that created all the code that controls all the traffic lights within the city of Dublin.

Update 2:

I collected all the data today from the intersection in downtown Dublin. I took measurements on the amount of cars during five cycles of the traffic lights and also took some times of the traffic light cycles.  Now I need to put this data in a spreadsheet, then use my data and the data collected by the city of Dublin on car frequently and speed, in the software modeling software to create a visual representation of the intersection. That will complete the active part of my project, and then all I will need to do is complete the capstone paper.

I still am researching which traffic modeling software will work the best f0r my purposes. However I have looked at a few options and the best candidates currently are SUMO, MATSim, and TRANSIMS. These are all opensource options that I believe have the features I am looking for I will download these options and see if they work for a complex single intersection, and if none of them do I will look for further options.

I am planning to have the conclusion blog post uploaded on Saturday night or Sunday morning. This means I need the intersection done by that time as well. After that I will write up my paper, and work on a poster design.

Update 1:

This update is a bit late because creating the code for this project is much greater difficulty then expected. I knew that it would be hard to create but my knowledge in this field of time based motion is so limited. I still want to finish this project, and am starting my capstone paper today, but how i get to my end project will most likely be changed from my original plan to code it all from scratch in Java. I am giving myself a little bit more time to try and create the model from scratch but if that fails, and it is likely given that a I have had to relearn some now that it is summer and that learning a new topic through emails and online methods is very hard.

Therefore if I am unable to create the model from scratch I will be using software / simulation packages already online. I believe that this will be the best way to complete the project as I will still be analyzing local data collected by the city, and with this extra time I can go out and collect data as well. I have already learned alot about how this software would be made but the fact is that creating traffic models from scratch with little knowledge is a very hard thing to complete in only a year.

In my next update I hope to have collected my data, and have a way to use that data in a model whether that is from my own software or from someone else’s. Tomorrow is when I am collecting the data on the intersection of Downtown Dublin. I choose this intersection because it is very busy, is a four way stop light intersection, and has multiple lanes in some directions. Below is a picture of this intersection in question.

For my capstone project, I am going to be making a model of a Dublin, Ohio intersection. This project is due in two weeks and there is much to complete still. I have found all the resources for my project. This includes examples of traffic simulations to use as a starting point of my code, and text of traffic flow modeling.

To insure that I will finish this project I am splitting my work into two phases. this week will be blog posts and the writing of the actual code. next week will be writing the paper and adding finishing touches to anything left to do.

Therefore, the basic method for this project is looking at examples of traffic simulations and taking features from many and combine them in an original way. I will also be using local data to help guide the movement of the cars. These addition features that I haven’t seen in traffic simulations much will add human error and hopeful a more realistic end project. I will then report my experience in the capstone paper.

In my next update I will have a basic draft of the code, and a finished list of all features. This update will be done uploaded by Wednesday. I understand that this is a very quick timeline now, but with the amount of research and resources I have from the past months, I know that this project has and will teach me a lot about a topic I didn’t know anything about.

Introduction

My capstone project focuses on data visualization regarding traffic flow, and the motion of individual cars. This project is easily scaled from one intersection to entire city. This will allow me to test my program throughout development and improve it before scaling up. Also if the time to create one intersection is great then the project can be capped easily there.

My project will have a product of a software that will show traffic on roads throughout a time frame. This traffic will be shown through actual cars if possible, not simple colorize of the road. A focus of the car’s motion will be on traffic lights/ stop signs, lane changes and congestion, as well as the movement of cars from location to location, as cars are used for actual transportation between physical places.

Timeline

• contact professors to be advisers
• research how to get started / ask questions
• collect timed data from websites ( on one intersection at first)
• load data into software ( most likely java)
• model the physical roads and traffic signs
• model a car object ( location (x,y), velocity (x,y), acceleration (x,y) )
• use data and object to model cars motion based on time
• deal with making car objects appear and disappear ( from cars arriving and debarking on journeys)
• deal with interactions between cars ( congestion)

• improve cars modeling
• add more roads and intersections to cars motion