Login

Project

#61 Real-time traffic monitoring and prediction for Cranberry Township


Principal Investigator
Sean Qian
Status
Completed
Start Date
Sept. 1, 2017
End Date
Dec. 31, 2018
Project Type
Research Applied
Grant Program
Private Funding
Grant Cycle
2017 Smart Mobility Challenge
Visibility
Public

Abstract

Cranberry Township is a Progressive Municipality that works to maintain traffic efficiency on its transportation networks. Cranberry Township’s unique Geographical location at the junction of Interstate 79 and the Pa 376 (Pa Turnpike) can pose interesting challenges in coordinated operations. 

Specifically, the Township operates a Coordinated Signal System that relies on historically Generated Signal Timings, coupled with real time technology to manage day to day operations on the local network.  Unfortunately, any scheduled or unscheduled events on the limited access highways can cause havoc with our efficient operation.

Together we are proposing to incorporate real time data inputs monitored from both social media and INRIX against historical INRIX data from these limited access highways to trigger predictions of traffic delays. These predictions could then be directed to several directives, such as dynamic message boards, smart phone applications, social media and text messages, to alert the public of the anticipated delay. Those predictions also alert the Cranberry traffic system of the issue to allow for adjustments to the operating traffic plan on the real-time basis.
    
Description
Task 1: Identify various data sources for in-depth data analytics
1.	Establish a refined GIS model for Cranberry Township based upon the SPC model. A stand-alone version of Township GIS with the following data is necessary for this study, which should include street names, street levels (highway, major arterials, minor streets, alleys, etc.), the number of lanes, and speed limit.
2.	Obtain incident data recorded by Cranberry Township Public Works and by the CMU Mobility Data Analytics Center (MAC)
3.	Obtain historical travel time data on two corridors near Cranberry Township (SR 19 and Route 228) from MAC
4.	Obtain traffic counts on local streets from Cranberry Township
5.	Obtain traffic counts on highways from PennDOT if available. 


Task 2:  Identify modeling scenarios in the region
We will work with the Cranberry Township to gather detailed information about the construction plans for roadway closures and any unplanned events that caused substantial traffic delays in the past two years, including geographical scope of the closure, time/duration of closure and lane closure configurations. We will also identify significant incidents that occurred on the arterials in the past two years. We plan to model traffic in the following tasks for each of the non-recurrent traffic scenarios in addition to recurrent traffic.  

Task 3:  Establish statistical models for predicting travel times for all road segments in Cranberry Township 
We will establish a statistical data mining model, such as deep learning networks or non-linear regression models, to capture the linkage between historical traffic data, real-time data and next-hour travel time for all road segments. For each of the scenarios identified in Task 2, we will calibrate the model separately for better prediction results. The application of this statistical model is to real-time predict travel delay and alert managers of delay beyond a pre-determined threshold on critical locations by real-time mining historical data and real-time INRIX/traffic counts data. The model will be intensively tested for the period of this project and adjusted according to results and comments from the township managers/engineers/operators. 
Task 4: Model the potential benefits of traffic mitigation plans
In addition, we examine the effectiveness of potential traffic management strategies. We will use a mesoscopic network analysis methodology to conduct this research. CMU Mobility Data Analytics Center uses a dynamic network analysis tool (MAC-POSTS) capable of estimating network-wide traffic impact of planned or unplanned incidents for any general networks consisting of freeway, arterials and local streets.MAC-POSTS can predict the effect of those hypothetical construction plans and/or incidents in high spatial and temporal resolution.
For each scenario identified in Task 2, we also propose travel demand management (TDM) strategies to mitigate congestion. TDM strategies include on-site detour strategies and online detour information system. First, an on-site detour strategy will be proposed to suggest efficient detour routes and optimal traffic diversion ratios. The optimal diversion can be achieved by deploying electric message signs in the right locations and with proper texts. We will also develop online-detour information system to dissimilate construction/incident information and suggest detour guidelines for travelers.  
The results of this research can be directly used to help manage future large-scale reconstruction projects and unplanned incidents: 1) its guidelines can streamline the development of construction plans, and 2) its models can be used to evaluate the travel delay, fuel consumption, and emissions of alternative construction/traffic management plans in a corridor network, thus facilitate the choice of the most effective plan for both closures and TDM. 
Timeline
Sep:  Task 1: Identify various data sources for in-depth data analytics
Oct: Task 2:  Identify modeling scenarios in the region
Nov-Jan: Task 3:  Establish statistical models for predicting travel times for all road segments in Cranberry Township 
Feb-Aug: Task 4: Model the potential benefits of traffic mitigation plans
Strategic Description / RD&T

    
Deployment Plan
We will work closely with the Cranberry Township throughout this project. While we focus on several particular applications (travel time of roadway in this project) to demonstrate the method and leverage our resources, the methodology can be broadly applicable and scalable to other cities or regions, and other metrics (such as transit travel time, parking occupancy, etc.). This generality will attract attentions from various groups interested in smart infrastructure, green design, and environmental policies. In the future, potential funding agencies/collaborators include the state DOT, the City of Pittsburgh, SPC, local governments, communities and foundations.
Expected Outcomes/Impacts
The desired results are a large-scale high-resolution statistical model built for the Cranberry Township in the Pittsburgh region that supports decision making for traffic operators and managers.  Performance metrics are the prediction accuracy obtained from model training and field test. We expect to achieve minimum 80% accuracy in terms of mean squared error for the travel time in the next hour. 
Expected Outputs

    
TRID


    

Individuals Involved

Email Name Affiliation Role Position
seanqian@cmu.edu Qian, Sean Carnegie Mellon University PI Faculty - Untenured, Tenure Track

Budget

Amount of UTC Funds Awarded
$60000.00
Total Project Budget (from all funding sources)
$60000.00

Documents

Type Name Uploaded
Publication Understanding and predicting highway travel time with spatio-temporal features of network traffic flow, weather and incidents March 25, 2018, 11:48 a.m.
Presentation Mobility Data Analytics March 25, 2018, 11:48 a.m.
Progress Report 61_Progress_Report_2018-03-30 March 25, 2018, 11:48 a.m.
Presentation Mobility Data Analytics Sept. 23, 2018, 7:40 p.m.
Progress Report 61_Progress_Report_2018-09-30 Sept. 23, 2018, 7:40 p.m.
Final Report 61-Final.pdf Feb. 12, 2019, 9:55 a.m.
Publication Understanding transit system performance using avl-apc data: An analytics platform with case studies for the pittsburgh region. Dec. 2, 2020, 9:09 a.m.
Publication Statistical inference of probabilistic origin-destination demand using day-to-day traffic data. Dec. 2, 2020, 9:09 a.m.
Publication Managing traffic with raffles. Dec. 2, 2020, 9:19 a.m.
Publication Learning to Recommend Signal Plans under Incidents with Real-Time Traffic Prediction. Dec. 2, 2020, 9:35 a.m.
Publication An Optimized Temporal-Spatial Gated Graph Convolution Network for Traffic Forecasting. Dec. 2, 2020, 9:37 a.m.

Match Sources

No match sources!

Partners

No partners!