Presentation is loading. Please wait.

Presentation is loading. Please wait.

Assessment and Refinement of Real-Time Travel Time Algorithms for Use in Practice January 10, 2007.

Similar presentations


Presentation on theme: "Assessment and Refinement of Real-Time Travel Time Algorithms for Use in Practice January 10, 2007."— Presentation transcript:

1 Assessment and Refinement of Real-Time Travel Time Algorithms for Use in Practice January 10, 2007

2 Agenda Other States’ Efforts (15 minutes) Data Collection & Analysis Update (20 minutes) OTREC Proposal (10 minutes) INRIX Follow-Up (5 minutes) Deliverables (5 minutes) Next Steps (5 minutes)

3 Best Practices – Other States..

4 Travel Time Status Alaska Hawaii Puerto Rico D.C. = Provide Travel Times (25) = Plans to Provide Travel Times (17) September 15, 2006

5 Methods… Two main approaches for generating travel times  In house Loop Detectors Video Detectors RTMS Toll Tags  Private providers Smartroute Systems Inrix Traffic.com

6 What others are doing…. In house  Milwaukee  San Antonio  Chicago  Atlanta  Nashville  Houston Private Providers  Boston  Miami  St. Louis  North Carolina  New Jersey

7 Milwaukee, WI Detector Spacing  0.25 miles in urban areas  2 miles in rural areas Data from detectors transmitted to Traffic Operations Center Freeway Traffic Management System (FTMS) Server  Travel Time = Known Distance/Average Speed Website updated every 3 minutes DMS signs updated every 1 minute No probe vehicle data; all detector derived travel times

8 San Antonio, TX Travel Times calculated from/to major interchanges Detectors  Loop Detectors  Video Detectors Point travel speeds used to calculate times from detector to detector based on distance between detectors Travel times displayed on mainlane DMS from 6:00 a.m. to 10:00 p.m., seven days of the week Travel times posted on TransGuide website use the same algorithm

9 Data Calculation Travel Time Calculation- - Segment travel speed is chosen as lower of upstream and downstream sensor speed - Point to point travel speed is summation of segment travel times Source: San Antonio TransGuide Travel Time Algorithm

10 Chicago, IL  IPASS Data Travel times from toll plaza to toll plaza Based on toll transponder data collected by ETC system > 1.5 million users on tollways Significant number of probe vehicles provide time stamp and location Travel times calculated using location and time stamp information High quality of data  RTMS Data  IDOT Loop detector data

11 Travel Time Estimation Toll network  Roadway segments  Bounded by ramps and plazas IPASS & RTMS data  Provide redundancy on certain segments  Travel time software allows choice of data DMS Travel Times  From all three sources (I-PASS, RTMS, Loops) GCM Webpage  Only I-PASS travel times

12 Houston, TX AVI Toll Transponders  Approximately 2 million transponders  Data collected at 232 reader stations and transmitted to Transtar TMC  Reader stations 2-3 miles apart Automated Travel Time Processor  Posts travel times to 81 DMS signs every 10 minutes (5:30 a.m. – 7:30 p.m.)  Some signs updated more frequently than others

13 Nashville, TN Technology  RTMS detectors at 0.25 mile spacing Speeds Ensure data quality by regular calibration  CCTV cameras Travel Time verification Data Collection & Processing  Average speed from RTMS every 2 minutes  Travel time calculation Average speed and distances between sensors  Travel Times 2-3 minute ranges Segments < 5 miles

14 Atlanta, GA Changeable Message Signs (CMS)  All major freeways  HOV CMS Information for express lane commuters  Automatic message generation  Travel Times between 6 a.m. – 9 p.m. Average speeds from Video Detection Cameras  Incident Messages Incident location Number of lanes affected

15 Technology VDS Cameras  Monitoring and Video Detection Cameras  Fixed black and white cameras  Placed along all major freeways  Provide volumes and speeds Source: http://www.georgia-navigator.com/about.shtml

16 San Francisco - Bay Area, CA Existing Caltrans System  Dual Loop Detectors Speeds New MTC System  Antennas to read FasTrak Toll Tags  Average Travel Times and speeds of all vehicles 511 System  Combination of data from both sources to calculate travel times

17 Seattle, WA Loop Detectors  Speeds  Travel Times Current Times  Speed of two adjacent stations is averaged  Distance/Average Speed = Travel Time Adjustment  Average Travel Times for each route by time of departure  Determination of future conditions based on historical data  Conditions worsen – adjustment is made  Conditions get better – no adjustment is made

18 Data Collection Dates: Thursday Dec 21, 2006; Tuesday Jan 9, 2007 Times: 7-9 AM, 4-6 PM both days Route: Drivers drove loop on I-5 South of Downtown (Tualatin-Sherwood – I-405) Method: Data collected using Garmin iQues – record lat/lon and speed every 3 seconds One iQue may be malfunctioing Two new iQues ordered – should be here soon Have around 15 runs for each segment of the I- 5 corridor (south of downtown)

19 Data Analysis Data storage and initial analysis procedures developed  Fix issues we had with previous analysis  Analysis must be automated Scripts written to extract data from csv files from iQues and insert data into the Portal database Database tables for storing information about  Raw data files (names, times, who collected, etc.)  Segments  Mapping between segments and stations and station lengths Easy to manipulate station lengths  Raw data  Travel times – ground truth and estimates

20 Two I-5 Segments I-5 NB Segment  I-5 NB from VMS N of Tualatin-Sherwood to I-405 junction  Milepost 290.9 to 299.8 I-5 SB Segment  I-5 SB from VMS N of Terwilliger to Tualatin Sherwood exit  Milepost 298 to 289.6 Also have start and ending points in lat/lon

21 Next: Segment Details I-5 NB first  Table of travel times  Timeseries speed  Plots of trajectories and speeds I-5 SB second  Table of travel times  Timeseries speed  Plots of trajectories and speeds

22 I-5 NB – VMS to I-405 – Travel Times RunDateTimeGround Truth Avg speed MidpointErrorError Pct 11-2-079:1816:1532.715:18-00:575.8% 21-3-079:1610:3649.2610:04-00:325.0% 31-4-078:4923:5421.8524:37 00:433.0% 412-21-067:1010:2749.3510:20-00:071.1% 512-21-067:4113:0040.4314:49 1:4914.0% 612-21-068:1011:0047.3311:40 00:40 6.1% 712-21-068:39 9:0657.65 9:45 00:39 7.1% 812-21-0616:3312:4540.9611:41-01:04 8.4% 912-21-0617:2123:0622.6222:37-00:29 2.1% Positive numbers indicate travel time was over-estimated (too high)

23 I-5NB – December 21, 2006 Timeseries – Speed – Shows level of congestion on I-5 NB on Dec 21, 2006

24 Run 5 – I-5 N – Trajectories Date: 12-21-06 Start: 7:41 Ground truth: 13:00 (780 sec) Midpoint3: 14:49 (889 sec)

25 Run 5 – I-5 N – Speeds

26 Speeds – Station 1014 – MP 297.33 Probe enters segmentProbe crosses detector

27 Run 9 – I-5 N – Trajectories Date: 12-21-06 Time: 17:21 Ground truth: 23:06 (1386 sec) Midpoint3: 22:37 (1357 sec)

28 Run 9 – I-5 N – Speeds

29 I-5 SB – VMS to I-205 – Travel Times RunDateTimeGround Truth Avg speed Midpoin t ErrorErrorpct 1012-21-067:28 8:4855.5 9:40 00:529.8% 1112-21-067:59 8:3057.3 9:18 00:489.4% 1212-21-068:27 8:0660. 9:31 01:2517.5% 1312-21-068:53 8:1559. 9:23 01:0813.7% 1412-21-0616:1316.4229.313:25- 03:1719.7% 1512-21-0616:5819:3624.919:09- 00:27 2.3% 1612-21-0617:5212:2739.413:19 00:52 7.0%

30 I-5SB – December 21, 2006

31 Run 13 – I-5 S – Trajectories Date: 12-21-06 Time: 8:53 Ground truth: 8:15 (495 sec) Midpoint3: 9:23 (563 sec)

32 Run 14 – I-5 S – Trajectories Date: 12-21-06 Time: 16:13 Ground truth: 13:00 (780 sec) Midpoint3: 14:49 (889 sec)

33 Run 14 – I-5 S – Speeds

34 Run 15 – I-5 S – Trajectories Date: 12-21-06 Time: 16:58 Ground truth: 19:36 Midpoint3: 19:09

35 Run 15 – I-5 S – Speeds

36 Data Collection – Next Steps Collect more data  Corridor? I-205? (217 still under construction) Incorporate old data into new analysis system Complete analysis

37 OTREC Proposal Issues  Scope Extension  Phasing/Budget

38 OTREC Proposal - Scope Current Scope  Assess and improve travel time algorithms Scope extension for OTREC  Travel Time Prediction  Use historical data to improve prediction of near-term travel times Should I leave work now or should I wait half an hour?  Longer-term prediction will also be incorporated I’m going to the airport Friday at 4:30 – how long should I allow?  Useful and difficult from a research perspective Comments:  Current scope probably not sufficient for OTREC  Think we need research component and student support to obtain OTREC funding

39 OTREC Proposal - Budget Have ~$57,000 in match Current project ends June 30, 2006 OTREC year one ends Sept 30, 2007  Difficult to spend full $57,000 by then Propose:  Phase 1 – OTREC Year 1 - $25,000 through Sept 30, 2007  Phase 2 – OTREC Year 2 - $32,000 through Jan 30, 2008 Budget breakdown similar to current budget  $5,000 data collection ($10,000 total)  Kristin 0.25 time  Sirisha + 1 other student

40 Deliverables Task 1 – ITS Data Fidelity – Submitted Task 2 – Ground Truth Data Collection – In Process Task 3 – Sensitivity Analysis – Submitted Task 4 – Algorithm Refinement Task 5 – Detailed Comparison of Algorithms Task 6 – Draft Final Report Task 7 – Final Report

41 Next Steps Additional Ground Truth Collection Analysis of Ground Truth Data Follow-up with Inrix Comments??


Download ppt "Assessment and Refinement of Real-Time Travel Time Algorithms for Use in Practice January 10, 2007."

Similar presentations


Ads by Google