1. Completing the Real-Time Traffic Picture Stanley E. Young, P.E. Ph.D. University of Maryland Center for Advanced Transportation Technology Traffax Inc.

2. Outline  Vehicle Probe Project  Where we have been, where we are now ….  Completing the Picture  Bringing in Volume Data – State Wide  Extending Real-Time to Arterial Networks  Its time for Arterial Management Systems …

3. I-95 Corridor Coalition  Maine to Florida  Road, Rail, and Water Transport  DOTs, Toll authorities, Public Safety, and other related organizations  Forum for management and operations issues  Volunteer organization  Goal: Improve Transportation System Performance  Multi-jurisdictional cooperation, >20 years

4. Vehicle Probe Project  Phase I (2008-2014)  First Probe-based Traffic System  Specifications-based, validated  Licensing - one buys, all share  Began 2.5K miles, grew to 40K  Travel time on signs, 511 systems, operational awareness, performance measures  Phase II (2014 forward)  All of the above  Better quality, less cost  Data market place (Multiple- vendors)  Emphasis on arterials and latency  42.5K and growing  Map-21 Performance Measures

5. Vehicle Probe Project  Phase I (2008-2014)  First Probe-based Traffic System  Specifications-based, validated  Licensing - one buys, all share  Began 2.5K miles, grew to 40K  Travel time on signs, 511 systems, operational awareness, performance measures  Phase II (2014 forward)  All of the above  Better quality, less cost  Data market place (Multiple- vendors)  Emphasis on arterials and latency  42.5K and growing  Map-21 Performance Measures

6. First Multi-Vendor Freeway Validation I-83 & I-81 Harrisburg, Oct 2014  PA-08  14 Segments  31.3 miles  Data collection  2300 to 2555 total hrs  71 to 80 hrs [0-30]  53 to 66 hrs [30-45]  AASE  2.1 to 4.1 mph [0-30]  3.1 to 5.8 mph [30-45] 6 May 07, 2015

7. PM Peak Hour (Oct 15-16, 2014) May 07, 2015 7

8. 8 Non-recurring Congestion Oct 13, 2014 10 AM to 7 PM

9. Integrating Real-Time Volume  Objective: Provide volume data in real-time on freeways and high-level arterials in a method similar to probe based speed and travel time data  Approach  Cooperative Research Initiative with Industry  Calibration/validation test bed  Focus group to refine product  Vendors develop, test, and report  Goal is to accelerate timeframe to viable real- time volume data feed  If interested contact rmjcar@umd.edurmjcar@umd.edu

10. Arterial Probe Data Quality Study 2013 – mid 2014 10 State / Set ID Road Number Road Name Validation Date Span # of Segments # of Through Lanes AADT Range (in 1000s) Length* (mile) # Signals / Density # of Access Points Median Barrier Speed Limit (mph) NJ-11 US-1 Trenton Fwy, Brunswick Pike Sep 10 - 24, 2013 102-433 - 9014.210 / 0.7112Yes55 NJ-42Black Horse Pike8225-5412.523 / 1.8260Yes45-50 US-130Burlington Pike1034214.328 / 2.0229Yes50 NJ-12 NJ-38Kaighn Ave. Nov 5-19, 2013 162-432-8024.544 / 1.8235Yes50 NJ-73 Palmyra Bridge Rd. 182-433-7423.941 / 1.7236Yes45-55 PA-05 US-1Lincoln Highway Dec 3 - 14, 2013 282 - 3+321 - 10030.62107 / 3.5178Yes40 - 50 US-322 Conchester Highway 61-222 - 3414.287 / 0.548No35 - 45 PA-06 PA-611Easton Rd Jan 9 - 22, 2014 102-418-316.721/ 3.1398NO40-45 PA-611Old York Rd81-221-307.326/ 3.56105Partial15-40 PA-611N Broad St162-417-329.6102/ 10.62161NO15-40 VA-07 VA-7 Leesburg Pike and Harry Byrd Hwy April 5-16, 2014 302-445-6030.557 / 1.9203Yes35-55 US-29 Lee Hwy (S Washington St) 4214-254.422 / 5114Partial30 VA-08 US-29Lee Hwy May 8-19, 2014 262-415-4531.9115/3.6287Partial35-50 MD-08MD-140 Reistertown Rd June 5-14, 2014 121 - 319-4410.540 / 3.8148No30-40 Baltimore Blvd62 - 440-5311.016/ 1.538YES45-55 9 Case Studies from 2013-14 Spans NJ through NC Test extent of probe data 15K AADT to 100K 2 – 12 lanes 0.5 to 10+ signals per mile Objective: Reference case studies April 30, 2015

11. Arterial Probe Data Recommendations 11 Probe data quality most correlated to signal density Increased volume aids probe data quality, but does not overcome issues resulting from high signal density Accuracy anticipated to improve with increased probe density and better processing April 30, 2015 Likely to have usable probe data Possibly usable probe data Likely not usable probe data <= 1 signals per mile AADT > 40000 Fully or Partially captures >75% slowdowns <= 2 signals per mile AADT 20K to 40K May Fail to capture > 25% of slowdowns Should be tested >=2 signals per mile Not recommended

12. Additional Insights 12  Not ready for Prime Time  Probe data usable on highest class arterials  Signal density < 1 per mile on average  Travel times are proportional to ground truth  May still miss some slowdowns, and may want to test  Consistent positive bias at low speeds  As probe data improves, delay will increase  Remaining Challenges  Severe queuing / multi-cycle delays  Optimistic bias with bi-modal traffic  Not sensitive to signal timing changes  Major disruptions problematic April 30, 2015

13. Roadmap for Arterial Management Systems

14. Technologies Enabling Arterial Management Systems Re-identification  Samples vehicle travel time (5% for BT)  Works best at corridor level  Independent of Signal System  Provides top-level user experience information High-Res Signal Data  Logs all actuation and phasing information  Works best at intersection level  Integrated with Signal System  Provides detailed intersection analysis Both enabled by consumer wireless communication and big data processing. Available Now – Multiple Vendors - Cost Effective Not one or the other… but both!

15. Emerging Arterial Performance Measures  Travel Time and Travel Time Reliability – based on sampled travel time sources  Initially re-identification data, later outsourced probe data as it matures, as well as connected vehicle data  Percent Arrivals on Green  Supported by methods such as Purdue Coordination Diagram tools  Split Failures (occurrences)  Related to GOR / ROR and similar analysis

16. Re-Identification Data (Bluetooth)  Uses a ID unique to a vehicle (MAC ID of a Bluetooth device inside vehicle)  An initial detector identifies when a vehicle enters a corridor by the vehicle’s ID  Another detector re- identifies the vehicle at the end of the corridor  Travel time/ speed can be directly calculated from the entry and exit time 16 Picture source: libelium.com Direct samples of Travel Time

17. Re-identification Data Fidelity

18. Statistical Performance Measures

19. High Resolution Signal Data  Logging of sensor and phase information  Data forwarded periodically to central server  Applications  Purdue Coordination Diagram  Red-Occupancy Ration / Green Occupancy Ratio  Volume / Demand Analysis (per movement)  Streamlined Maintenance Picture Source: FHWA

20. Sample Metric - Intersection Purdue Coordination Diagram

21. Sample Metric - Intersection Movement Capacity Analysis (ROR – GOR)

22. Current State of Arterial Management Systems (AMS)

23. Benefits  Created a language between traffic engineers, planners, operations, and management to establish goals, measure performance, and manage the system  Link performance to budget/funding  Systematic approach  Long term performance tracking  Better utilization of professional staff  Organizational maturity

24. Real-Time Arterial Performance  Stuff from Chapter 1

25. Travel-Time Visualization  Picture of Cascading CFD  Picture of ROR and GOR

26. Intersection Impact – ROR & GOR

27. Thank You! Stanley E. Young, P.E. Ph.D. University of Maryland Center for Advanced Transportation Technology 5000 College Ave, Bldg 806, #2203 College Park, 20742 Mobile 301-792-8180 seyoung@umd.edu