Comparative Analysis Of Various Travel Time Estimation Algorithms To Ground Truth Data Using...
-
Upload
jennifer-fleming -
Category
Documents
-
view
218 -
download
2
Transcript of Comparative Analysis Of Various Travel Time Estimation Algorithms To Ground Truth Data Using...
Comparative Analysis Of Various Travel Time Estimation Algorithms To Ground Truth Data Using Archived DataChristopher M. Monsere
Research Assistant ProfessorDept of Civil and Environmental EngineeringPortland State University
Sirisha Kothuri, Kristin Tufte, Robert L. Bertini,
School of Urban & Public AffairsPortland State University
Incorporating Freight Performance Measures in a Regional Transportation Data Archive
Christopher M. MonsereRobert L. Bertini Zachary Horowitz Kristin A. Tufte
Department of Civil & Environmental EngineeringIntelligent Transportation Systems Laboratory
Portland State University
NATMECJune 7, 2006
Minneapolis, Minnesota
Photo J. Fischer
2
Outline
Brief description of existing archiveExisting performance measurementPossible freight data sourcesSome resultsNext steps
3
PORTALPSU Designated as Regional Archive Center
PSU participates in regional ITS committee—TransPort
PSU designated as regional center for collecting, coordinating and disseminating variable sources of transportation data and derived performance measures.
Bi-state: Oregon and Washington
May expand statewide.
4
PORTAL Architecture Regional ITS Data Sources
98 CCTV Cameras 19 Variable Message
Signs (VMS) 485 Inductive Loop
Detectors 135 Ramp Meters Weather data TriMet Automatic Vehicle
Location (AVL) System and Bus Dispatch System (BDS)
Extensive Fiber Optics Network
5
Data FundamentalsWhat Data Do We Collect?
20-second Intervals Freeway
Mainline Count Occupancy Time mean speed
20-second Intervals Freeway On-ramps Count
Hourly @PDX Temperature Precipitation
Every four hours @PDX Weather descriptor (e.g., clear, mist,
light rain, rain, fog, light snow)
6
Data FundamentalsPerformance Measures We Compute
Segment Length
Vehicle Miles Traveled (VMT) = Count Segment Length
Travel Time = Segment Length Speed
Free Flow Travel Time = Segment Length Free Flow Speed
Vehicle Hours Traveled = Count Travel Time
Delay = Travel Time − Free Flow Travel Time
7
Data Analysis and VisualizationHomepage
8
Data Analysis and VisualizationContour Plots: Speed
9
Data Quality Popup
10
Data Analysis and VisualizationTime Series Plots: Volume
11
Data Analysis and VisualizationGrouped Data Plots: Speed
12
Performance Reports
13
Monthly Reports
14
Mapping and Spatial AnalysisSpeed by Month
Average Evening Peak Speed (5PM-6PM)
July 2005 Dec 2005
15
Mapping and Spatial AnalysisTravel Time Reliability
Point to Point Off-Peak Travel Time Reliability (I-5 N)
Point to Point Peak-Hour Travel Time Reliability (I-5 N)
16
Incident Contour Plot
Incident on I-205N at the off ramp for Hwy 212/214. A log truck rear-ended a truck carrying nursery stock; two cars also involved. Incident lasted just over 4 hours.
11/15/2005 Northbound I-205
17
Incident Contour Plot
11/15/2005 Southbound I-205
Incident on I-205 Southbound.
One lane closed.
18
PORTALPotential Freight Data Sources
Vehicle classificationContinuous - short and long (detector stations)Fixed classification sites
Weigh-in-motion Truck monitoring (AVI)
CorridorPoint
Spot vehicle classifications
Port of Portland / METRO / ODOT / WsDOT truck count effortOther traffic study counts
19
Potential Freight Data Weigh-in-motion and Truck AVI
23 sites 1.5 years of data stored, not yet archived
Researching its usefulness
AVI tag matching projectMerging with other data
20
Potential Freight Data Spot vehicle classifications
21
Potential Freight Data Vehicle classification
Current firmware/software not configured to detect length PSU MTIP grant to help fix
Instead, uses vehicle classification algorithm for single loop Wang and Nihan– occupancy, count, speed estimation factor– a number of assumptions
22
Potential Freight Data Vehicle classification
23
Potential Freight Data Vehicle classification
Compare Wang and Nihan estimates Manual counts Video processed
Site selection (4) Existing CCTV and loops in same field of view Detector quality Trucks, no congestion Ability to request PTZ
Manual count Short (<39 ft), Long (>39 ft) Timer based count of video
Autoscope 8.1 RackVision analysis of DVR data
24
Potential Freight Data Vehicle classification
25
Potential Freight Data Vehicle classification
26
Next steps and Concluding Remarks
Continue to explore adding freight data to archive
Vehicle classification WIM data AVI tag
Data quality?? Web interface is expanding use of
archived data Increasing awareness of the value of
these systems Provides decision support for
transportation officials in the region
27
Acknowledgments PORTAL Team: Kristin Tufte, James Rucker,
Spicer Matthews, Jessica Potter, Sue Ahn, Sirisha Kothuri, Andy Delcambre, Tim Welch, Steve Hansen, Andy Rodriguez, Andrew Byrd
National Science Foundation Oregon Department of Transportation City of Portland TriMet Portland State University Oregon Engineering and Technology Industry
Council
Visit PORTAL online at:http://portal.its.pdx.edu
28
Thank You!
Questions?