���

Leading Pedestrian Interval (LPI) Implementation Guidelines in Phoenix, Arizona: A Data Driven Approach

October 10^th, 2024

2024 ITS Arizona Conference

Mesa, AZ

Motivation / Process

• City of Phoenix desired public facing LPI implementation guidance
• Part of a larger project; also looked at SOP recommendations for pedestrian scramble, among other items
• Process
• Literature Review
• Site identification through crash / volume analyses
• Data collection / reduction (with / without LPI)
• Data analysis / modeling
• Guideline development

​

2

Site Identification

• Priority Intersections Identification

​

• Pedestrian crash data from 2016 through 2022
• Pedestrian – Turning Vehicle crashes
• Ranked from highest to lowest crash frequency
• Top 85 intersections, were separated into three different tiers

3

Selected Intersections for Volume Analysis

Tier 1 & Tier 2 Intersections

Site Identification

4

• High Exposure Crosswalks Selection through Volume Analysis

​

• Volume Counts: 12:00 PM to 1:00 PM and 5:00 PM to 6:00 PM
• Timing Plans to determine left turn phasing (FYA, protected only, protected-permitted, or permitted)

Conflicting Vehicles

CrossProduct of Washington/3rd

Site Identification

5

• Crosswalks Selection Criteria
• High value of CrossProduct
• No more than two crosswalks from any one intersection
• Two legs at Washington/3rd due to their high ped volumes and agency preference

​

​

6

Selected Crosswalks for Data Collection

7

Video Data Collection

• Summertime was avoided for data collection
• Before Phase- Nov 8, 23 to Nov 28, 23
• After Phase- Feb 23, 24 to March 05, 24
• 'New Traffic Pattern Ahead' signs were installed with LPI
• Data Collection- 7AM to 5PM, 10 hours each day
• Total 160 hours of video collected

​

• Indirect safety measures - assess potential risks and safety issues without relying solely on crash data
• Post Encroachment Time (PET) is popular tool for assessing conflict frequency and severity
• PET measures the proximity of two road users to occupying the same space at the same time, thus indicating the severity of a potential conflict.

​

​

8

Surrogate Measure for Assessing Safety

Concept of PET

9

Example Interaction

10

Video Data Reduction

• Primarily reduced by a group of four students
• A set of instructions was provided to the team
• Trial data reduction was performed to ensure consistency

​

​

11

Summary of Observed Ped-Veh Conflicts

• Total Conflicts (PET ≤ 5 sec)
• Before LPI – 1694 Conflicts
• After LPI – 1334 Conflicts

​

Pedestrian and Vehicle Characteristics during Interactions

12

• Mean Vehicle Speed slightly reduced after LPI
• Mean Pedestrian Speed slightly increased after LPI

​

13

Conflicts by PET-Veh Speed Severity

• Most Severe (1) - lower PET (≤ 1.5 sec) and high vehicle speed (≥ 11.5 mph) Conflicts
• Least Severe (6) - higher PET (> 3.5-5 sec) and low vehicle speed (< 11.5 mph)

Other Collected Factors

14

• Driver Violations
• Driver Evasive Actions
• Pedestrian Distractions
• Pedestrian Crosswalk Violation
• Pedestrian Evasive Action

• Pedestrian Group Size
• Pedestrian Gender
• Pedestrian Age
• Pedestrian with Additional Mobility Devices

Compliance with Traffic Rules

Pedestrian Demographic and Mobility Characteristics

Analysis of Pedestrian-Vehicle Conflict Frequencies

• Negative Binomial Model - non-negative conflict data

15

• A series of NB models were developed
• All Reduced Conflicts (PET ≤ 5 sec)
• High Severity Conflicts (PET ≤ 1.5 sec)
• High Severity High-Speed Conflicts

​

• With different combinations of turning vehicle volume
• All Turning Vehicle - Protected and Permitted Left Turns and Right Turns
• Conflicting Turning Vehicle - Permitted Left Turns and Right Turns
• Permitted Left Turns and Right Turns Separately

​

​

Conflict Frequency Models

17

C

Model Predicted Counts of (PET ≤ 1. 5sec) Conflicts

Predicted Conflict Equation

Npredicted_conflicts =exp[ -1.357 +(-0.673*LPI_indicator)+ (0.004*TurningVehVolume) +(0.012*PedVolume)]

Conflicting Turning Vehicle = Right Turns + Left Turns

Summary of Conflict Analysis

18

• Analyzed conflicts at 8 Phoenix crosswalks, focusing on PET ≤ 5 seconds.

​

• Significant reduction in total conflicts and severity after LPI implementation.

​

• NB regression models predicted 10-15% reduction in total conflicts and ~45-50% reduction in severe conflicts.

​

• Ordered logit model also confirmed significant reduction in high and medium severity conflicts.

​

• Factors like vehicle direction, driver violations, evasive actions and pedestrian group size influenced conflict severity

​

​

​

​

​

​

Crash Analysis Results

19

LPI Guidance Overview

20

LPI Guidance: Data Collection

21

LPI Guidance: Worksheet, Intersection

22

LPI Guidance: Worksheet, Crosswalk

23

LPI Guidance: Suitability Matrix

24

Optional

Minor safety benefit expected

Considered

Moderate safety benefit expected

Recommended

High safety benefit expected

​

​

​

​

​

​

LPI Guidance: Other Items

25

1. If there is no adjacent crosswalk, use 0-4 for Optional, 5-10 for Considered, and 11-15 for Recommended.
2. If not already installed, APS units should be used with all LPI implementations.
3. Due to the need for the voice prompts on APS units to be consistent to cross a given street, an LPI must be installed in pairs such that both crosswalks across a roadway operate with an LPI. Additionally, the ‘WALK’ indication must start at the same time on both crosswalks.
4. At locations with LPIs, left turn protection should lag the through movement only. Leading left turn protection for a movement across an LPI crosswalk can result in vehicles turning into the LPI.
5. For locations with a Flashing Yellow Arrow, the indication shall remain red until the LPI ends.
6. The guidance provided in this work is based on a 5-second LPI. A shorter or longer LPI duration can be used, based on engineer judgment and MUTCD guidance.
7. On coordinated facilities, the offsets of the adjacent intersections should be adjusted such the vehicle platoons do not arrive during the LPI.
8. Upon implementation of an LPI, signage such as ‘Traffic Control Change’ or ‘New Pedestrian Head Start’ may be implemented to inform drivers of the traffic control change.
9. To improve vehicular compliance with the LPI, the following items may be implemented:
1. Install visors to shield signal heads perpendicular to the LPI crosswalks so that drivers cannot anticipate the green interval. This is especially useful for skewed locations but can be an option at all locations.
2. No Right Turn on Red should be considered on approaches parallel to the LPI crosswalk, as the literature shows it increases the effectiveness of the LPI.
3. A ‘Turning Vehicles Yield to Pedestrians’ sign (R10-15) may be used to increase awareness of pedestrians to drivers.

​

​

​

​

​

​

LPI Guidance: Example

26

Note: Values for East and South Crosswalks assumed for example purposes

​

​

​

​

​

​

Limitation and Direction for Future Research

27

• Crash Analysis

​

• Conducted at the intersection level, not individual crosswalk level.
• Absence of observed pedestrian volumes in the crash analysis.

​

​

​

​

​

• Conflict Analysis

​

• Manual data reduction may introduce counting or measurement errors.
• Direct sunlight on or near traffic lights caused visibility issues; traffic light status assumed based on surrounding traffic.

​

​

​

​

​

Thank You.

Questions/Discussion?

Edward Smaglik edward.smaglik@nau.edu

Brendan J Russo brendan.russo@nau.edu

Steven Gehrke steven.gehrke@nau.edu

​

28