Top Banner
Leveraging Signal Infrastructure for Non- Motorized Counts in a Statewide Program: A Pilot Study Paper #15-5168 Presenter: Bryan Blanc Graduate Research Assistant Department of Civil and Environmental Engineering Portland State University 94 th Transportation Research Board Annual Meeting January 14 th , 2015 1
38

Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Jul 16, 2015

Download

Documents

TREC at PSU
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Paper #15-5168

Presenter: Bryan BlancGraduate Research AssistantDepartment of Civil and Environmental EngineeringPortland State University

94th Transportation Research Board Annual MeetingJanuary 14th, 2015

1

Page 2: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Paper #15-5168

Research Team: Miguel Figliozzi

Christopher Monsere

Krista Nordback

Pam Johnson

Bryan Blanc

2

Page 3: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Research project goals

Pilot study

Pedestrian results

Bicycle results

Overall Performance of 2070 controllers and inductive loops

Lessons learned and recommendations

Presentation Outline

3

Page 4: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Research Project Goals

• SPR # 754: Design and Implementation of Pedestrian and Bicycle Specific Data Collection Methods in Oregon

Review best practices regarding statewide non-motorized data collection programs and data collection technologies

Provide guidelines regarding location and data collection procedures

Perform a pilot study

Agenda – Findings – Discussion – Next Steps 4

Page 5: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Pilot Study

Evaluate 2070 traffic signal controllers and inductive loops as methods for pedestrian and bicycle counting on a typical ODOT facility

Apply factors to estimate bicycle and pedestrian Average Annual Daily Traffic (AADT)

5

Page 6: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Site Selection Criteria

Adequate volume of pedestrians and bicyclists

On ODOT facilities

2070 signal controller

Pedestrian push-button to request a pedestrian phase (for counting pedestrians)

Bicycle lane inductive loops (for counting bicycles)

Staff support/recommendation

6

Page 7: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Pilot Study Site

OR-99W & Hall Boulevard, Tigard

99W AADT ≈ 41,000

Hall AADT ≈ 16,000

7

Page 8: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Pedestrian Data

8

Page 9: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

• PSU Data Collection

Preliminary Site Prep

• Counting Cars Video Recorder

• 3 cameras

• 24 hours

• 8/29 9 AM – 8/30 9 AM, 2013

9

Page 10: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Pedestrian volumes – video counts

0

10

20

30

40

50

60

9:00 AM 12:00 PM 3:00 PM 6:00 PM 9:00 PM 12:00 AM 3:00 AM 6:00 AM

Ped

estr

ian

Vo

lum

e

9AM August 29 - 9AM August 30, 2013

10

Page 11: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Pedestrian volumes – video counts

Peak hours: 12 PM to 6 PM43% of all volume

11

Page 12: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

75

92

68

108

4

12

8

33

1

4

3

12

6

1

10 3

0% 20% 40% 60% 80% 100%

North

South

East

West

Percentage of Observations

Cro

ssw

alk

1 2 3 4 5Pedestrian Group Size

Pedestrian Activity by Crosswalk and Group Size

Total observations: 440 groups (1 or more per group)Total: 596 pedestrians

12

Page 13: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

R² = 0.75

0

5

10

15

0 5 10 15

VID

EO C

OU

NTS

LOGGED PEDESTRIAN PHASES

R² = 0.70

0

10

20

30

0 10 20 30

VID

EO C

OU

NTS

LOGGED PEDESTRIAN PHASES

R² = 0.87

0

5

10

15

0 5 10 15

VID

EO C

OU

NTS

LOGGED PEDESTRIAN PHASES

Actual vs. Logged Volumes by Crosswalk

R² = 0.74

0

5

10

15

0 5 10 15

VID

EO C

OU

NTS

LOGGED PEDESTRIAN PHASES

North Crosswalk South Crosswalk

West CrosswalkEast Crosswalk

North Crosswalk South Crosswalk

West CrosswalkEast Crosswalk

13

Page 14: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Potential sources of error

Pedestrian groups: phases vs. actual number of pedestrians

Some pedestrians push buttons in both directions

Confusion/error

Pedestrians have a long wait time

14

Page 15: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Potential sources of error

Some pedestrian phases are called by bicyclists (6%)

Bicycles on the sidewalk

Perceived safety

Confusion or lack of understanding

Long wait time

15

Page 16: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Overall Accuracy

482 pedestrian phases vs. 440 groups

91% accuracy (for groups of pedestrians)

482 Pedestrian phases vs. 596 people using the intersection

81% accuracy (for pedestrians)

Factor: 1.24 people per pedestrian phase

Strong correlations (R2>0.70) when broken down by crosswalk and hour

16

Page 17: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Overall Analysis: Pedestrians

Importance of video analysis

Demographic info and bike/pedestrian behavior

Counting pedestrian phases is a promising and

cost-effective method for AADT estimation!

…but it is necessary to increase the number of

sites/analyses before generalizing results

Different sites, land use, traffic impacts, etc.

17

Page 18: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Bicycle Data

18

Page 19: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

• PSU Data Collection

Preliminary Site Prep

• Counting Cars Video Recorder

• 3 cameras

• 24 hours

• 8/29 9 AM – 8/30 9 AM, 2013

19

Page 20: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Bicycle Volumes: video data

0

5

10

15

20

25

9:00 AM 12:00 PM 3:00 PM 6:00 PM 9:00 PM 12:00 AM 3:00 AM 6:00 AM

Bic

ycle

Vo

lum

e

August 29, 9AM - August 30 9AM, 2013

20

Page 21: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Bicycle Volumes: 51% in bike line

21

0

5

10

15

20

25

9:00 AM 12:00 PM 3:00 PM 6:00 PM 9:00 PM 12:00 AM 3:00 AM 6:00 AM

Bic

ycle

Vo

lum

e

August 29, 9AM - August 30 9AM, 2013

Sidewalk Bike Volume Bike Lane Volume

Page 22: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Total bicycle volume: 190

Bicycles in bike lane: 97 (51%)

Bicyclists using pedestrian push buttons: 30 (16%)

Video Data Summary

22

Page 23: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Bike Loop Locations

23

Page 24: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Southbound Hall Boulevard to Tigard Library

Pilot Study

≈50 ft.

• Loop at stop bar

• Second loop approximately 50 feet from stop bar

• Loops in series, counted twice in 2070 data

• Note worn bike lane stripe !!!

24

Page 25: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Eastbound- 99W to Portland

Pilot Study

• Right turn pocket

• Single Loop before turn pocket

25

Page 26: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Loop Accuracy: Overcounting

%Error =#𝑙𝑜𝑜𝑝 − #𝑣𝑖𝑑𝑒𝑜

#𝑣𝑖𝑑𝑒𝑜

NB to Beaverton : 1474 %SB to Tigard: 1169 %EB to Portland: 5413 %WB to Sherwood: 2180 %

Eastbound- 99W to Portland

26

Page 27: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

• Getting the right inductive loop sensitivity is important to obtain accurate counts

• Sensitivity was subsequently lowered: only switches 2 and 8 set to ON (1 and 4 off)

• We recorded another 24 hour video session from 9 AM 10/24 to 9 AM 10/25

Analysis- Bicycle Loops

27

Page 28: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Loop Accuracy % Error =#𝑙𝑜𝑜𝑝 − #𝑣𝑖𝑑𝑒𝑜

#𝑣𝑖𝑑𝑒𝑜

7 % * 89% *2430 %61 %

Before

NB to Beaverton : 1474 %SB to Tigard: 1169 %EB to Portland: 5413 %WB to Sherwood: 2180 %

After

Accuracy did improve, but there may be a high correlation between vehicle traffic and bicycle detections– especially with high right turning volumes

* Based on 32 and 66 detections (see previous slide, the sum of two loops), the actual number of vehicles or bicycles that were detected at each bicycle loop on Hall is not known

28

Sensitivity Change:

Page 29: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

• Optimizing sensitivity is important to obtain accurate counts. Testing is necessary.

• The location of loops in relation to right turn movements is very important (EB to Portland very inaccurate)

• Right turn volume higher for Hall NB approach (less accuracy in counting bikes than Hall SB approach)

• Loops wired in series added difficulties in determining whether a bicycle was detected

Summary: Bicycle Loops

29

Page 30: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Overall Analysis: Bicycles

Importance of video analysis

Demographic/purpose info and bicycle behavior

Overall low accuracy in this study

Test other loop configurations

Analyze other intersections with better loop location

30

Page 31: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Overall Conclusions

Promising results: pedestrian counts utilizing pedestrian phases and simple factors

Not so good: counting bicycles utilizing inductive loops (many problems as explained)

Many more pedestrians than expected!

More research is needed to generalize the results

Another ODOT sponsored project is underway

31

Page 32: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Acknowledgements

• Technical Advisory Committee• Tiffany Slauter, ODOT Region 1 Signal Manager• ODOT TMS

• Don Crownover• Steve Chance, ODOT Traffic Data Analyst• Jacob Carringer (video setup)• Raymond Herrera (tubes installation)

• PSU Graduate Students• Sam Thompson• Adam Moore

• Bruce Moody : video documenting and interviewing

32

Page 33: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Questions?

33

https://pbs.twimg.com/media/BVGwDQLCEAAASMl.jpg

Page 34: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Pilot Study Land Use

34

Page 35: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Northbound Hall Boulevard to Beaverton

Pilot Study

• Loop at stop bar

• Second loop approximately 50 feet from stop bar

• Loops in Series, counted twice in 2070 data

35

Page 36: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Westbound - 99W to Sherwood

Pilot Study

• Single Bicycle Loop

36

Page 37: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Loop Accuracy

32*66*

25345

Video Counts

NB to Beaverton : 30SB to Tigard: 35EB to Portland: 10WB to Sherwood: 28

Loop Detections

10 hours with highest volumes: 6 AM to 11 AM and 2 PM to 7 PM

*32 and 66 are the sum of two loops; the actual number of vehicles or bicycles that were detected at each loop on Hall is not known

37

Page 38: Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study

Acknowledgements

Technical Advisory Committee• Lyn Cornell, ODOT Research Coordinator• Sheila Lyons, ODOT Bicycle & Pedestrian Specialist • Sue Geniesse, ODOT TGM Sr Planner • Gary Obery, ODOT Alt Mode Traffic Engineer • Julie Yip, ODOT Safety Div• Steve Lindland, ODOT Roadway Engineer Manager • Bruce Moody, FHWA

38