Evaluation of Pedestrian Hybrid Beacons and Rapid Flashing Beacons

CHAPTER 6. pHB STUDY

INTRODUCTION

This chapter describes the methodology and results from a study that examined driver and pedestrian behavior at PHBs. The PHB, or HAWK as it is known in Tucson, AZ, is a traffic control device used at pedestrian crossings. The crossing typically has the crosswalk across only one of the major road approaches. The PHB's vehicular display faces are typically located on mast arms over the major approaches to an intersection and in some locations on the roadside. An example is shown in figure 53 for an installation in Tucson, AZ, and in figure 54 for an installation in Austin, TX. The face of the PHB consists of two red indications above a single yellow indication. It rests in a dark mode, but when activated by a pedestrian, it first displays a few seconds of flashing yellow followed by a steady yellow change interval and then displays a steady red indication to drivers, which creates a gap for pedestrians to cross the major roadway. During the flashing pedestrian clearance interval, the PHB displays an alternating flashing red indication to allow drivers to proceed after stopping if the pedestrians have cleared their half of the roadway, thereby reducing vehicle delays.

Figure 53. Photo. Example of PHB installation in Tucson, AZ.

Figure 54. Photo. Example of PHBs being used in Austin, TX.

The PHB has shown great potential for improving pedestrian safety; however, questions remain regarding under what roadway conditions—such as crossing distance (i.e., number of lanes) and posted speed limit—should it be considered for use.^(26,27) In addition, there are questions about the device's operations. For example, a current topic of discussion within the profession is the way drivers treat a PHB when it is dark. PHBs dwell in a dark mode for drivers until activated by a pedestrian. A concern among some is that drivers will see a dark PHB and treat it as a Stop sign, similar to the required behavior for a dark traffic signal that has experienced a power outage.

The STC of the NCUTCD assists in developing language for chapter 4 of the MUTCD.⁽¹⁾ It is interested in research and/or assistance in refining material on the PHB. The PHB was first included in the 2009 MUTCD, which discusses the design and operations of the device along with guidance for installation categorized by low speed (roadways where speeds are 35 mi/h or less) and high speed (roadways where speeds are more than 35 mi/h).⁽¹⁾ The 2009 MUTCD also indicates that the PHB "...should be installed at least 100 ft from side streets or driveways that are controlled by Stop or Yield signs"⁽¹⁾ (pg. 449) In 2011, the STC recommended to remove that statement because it was a significant change from what was reviewed and approved by the National Committee in 2007 and what was proposed in the Notice of Proposed Amendment (NPA) for the 2009 MUTCD.⁽⁵⁰⁾ The statement was added to the PHB 2009 MUTCD discussion just prior to publication. The STC provided the following concerns with the 100-ft distance (with additional details added by this study's research team based on reviewer comments):

• The result of the added 100-ft guidance, if followed, is that these beacons could not be used at unsignalized intersections or driveways. The NPA language did not include any limitations (either standard or guidance) on the locations for use of the PHB; therefore, the 100-ft change was not subject to public review and comment.
  
  
• The 100-ft offset listed in the guidance is not supported by research or experimentation with this device. Most sites used for experimentation when the PHB was being tested were intersection or driveway locations which were the natural crossing locations. Therefore, the typical use of the device as tested, which ultimately proved to be successful, is recommended against in the 2009 MUTCD.
  
  
• All of the sites included in the FHWA study that evaluated the safety effectiveness of these devices were at stop-controlled intersections or major driveways.⁽²⁷⁾ The study was performed just prior to the publication of the 2009 MUTCD.
  
  
• The 100-ft guidance, if followed, causes increased mobility difficulties and discomfort for pedestrians with disabilities and forces all pedestrians to experience increased inconvenience if they must divert away from their desired crossing location at an intersection or driveway to a different crossing point located 100 ft or more away which would likely lead to 200 ft or more out-of-way travel. If the PHB is not placed at the natural crossing locations, it is likely it will not be used by most pedestrians, and their value as a safety device could be compromised.

Because of the questions being asked regarding driver and pedestrian behaviors with PHBs, FHWA sponsored a study to record behaviors at existing sites.

Study Objective

The objective of this study was to determine actual driver and pedestrian behaviors at locations with a PHB.

STUDY SITES

Through existing contacts and research team knowledge along with responses to requests, the research team compiled a preliminary list of PHB locations. Data for key variables (posted speed limit, number of through lanes, and the type of median treatment) were gathered and added to the list for the PHBs in communities with multiple installations. Pedestrian crossings on higher speed roadways and with wider crossings have historically experienced lower driver yielding, so posted speed limit and crossing distance (as reflected by number of lanes) were selected as key variables. The goal was to have at least eight sites with higher posted speed limits (defined for this study as being 40 mi/h or higher) and four sites with lower posted speed limits (defined for this study as being 35 mi/h or lower). The presence of a median can provide refuge for a crossing, which may affect the measures of effectiveness considered for this study, so it was also included in the original study matrix. Because of efficiencies in data collection, data were collected for a total of 20 sites. Roadway and traffic characteristics for the sites are listed in table 61.

NA = Not applicable.
^aSite name is denoted as AA-XXX, where AA represents the two-letter city code and XXX represents the number assigned to the site.
^bNumber of pedestrians per hour did not include any research team member crossings. They were observed during data collection (typically over a 4-h daytime period).
^cPHB is located within coordinated corridor where the timing of when the PHB is active is influenced by the nearby coordinated corridor.
^dFor midblock roadway configuration, the number in parentheses shows the distance (ft) measured from center of crossing to center of nearest driveway/intersection to the nearest intersection or major driveway.

The cities of Tucson, AZ and, Austin, TX, had the greatest variety in site characteristics of interest to this project and were selected for the study. Differences in practices between the two cities include the following:

• The Tucson, AZ, PHB faces had back plates with yellow reflective borders (see figure 53), while the Austin, TX, PHB faces did not have back plates (see figure 54).
  
  
• The signs used at most of the Tucson, AZ, sites included the CROSSWALK STOP ON RED (symbolic circular red) (R10-23) sign (see figure 55) and an internally illuminated PEDESTRIAN CROSSING or CROSSWALK sign (see figure 56). The sign used at the crossing for the Austin, TX, sites is shown in figure 57. This sign was selected to help educate drivers regarding appropriate behavior during the flashing red. Recently, FHWA has received numerous inquiries regarding how to address comprehension issues with the flashing red phase and is now recommending that if an alternative legend to the R10-23 sign is used, that it be the sign shown in figure 58.
  
  
• In advance of the crossing, Tucson, AZ, frequently installed a pedestrian crossing warning sign (W11-2) (see figure 59). School crossing signs were used at school sites.
  
  
• Austin, TX, included the STOP HERE ON RED (R10-6, R10-6a) sign at the stop line.
  
  
• The red clearance time (i.e., the elapsed time between start of the vehicular steady red indication and start of the pedestrian walk indication) was 1s at the Tucson, AZ, sites and 2 s at the Austin, TX, sites.
  
  
• The steady red interval was 8 s for Tucson, AZ, sites and ranged between 9 and 12 s for the Austin, TX, sites.
  
  
• When the Tucson, AZ, sites had a median, a PHB face and a CROSSWALK STOP ON RED (symbolic circular red) (R10-23) sign was frequently included on a post in the median.

Figure 55. Photo. Example of sign used in Tucson, AZ.

Figure 56. Photo. Example of internally illuminated sign used in Tucson, AZ.

Figure 57. Photo. Sign used in Austin, TX.

The currently preferred format for the type of sign shown in figure 57 is shown in figure 58.

Figure 58. Photo. Sign recommended by FHWA to address comprehension issues with the flashing red phase.

Figure 59. Photo. Example of advance warning sign used in Tucson, AZ.

The crosswalk markings were always located on only one side of the intersection. The PHBs had between 3 and 4 s of flashing yellow and between 3 and 4 s of steady yellow, consistent with city policies regarding clearance intervals at signalized intersections. For both cities, the flashing red duration varied based on the site's crossing width and ranged from 15 to 29s.

DATA COLLECTION AND REDUCTION

Data using a multiple video camera setup were collected in November 2014 for the Austin, TX, sites and in February 2015 for the Tucson, AZ, sites. All observations were collected during daytime dry weather conditions between 6:30 a.m. and 6:30 p.m. The observers and the video recording device were placed to be inconspicuous from the pedestrians, bicyclists, and motorists. The goal was to record a minimum of 50 pedestrian crossing events or 4 h of data (the smaller of the two) at each location, where each crossing event consisted of one or more pedestrian(s) crossing the entire width of the street. If it appeared that fewer than 50pedestrian crossing events would occur within the 4-h block, research team members would cross the street to increase the sample size of pedestrian crossings. Additional information on the staged pedestrian protocol is available in Characteristics of Texas Pedestrian Crashes and Evaluation of Driver Yielding at Pedestrian Treatments.⁽³⁵⁾ The research team members sought to complete data collection efforts at two sites per day, accounting for travel time between sites and the need to notify local stakeholders (e.g., school personnel) of their activity. Hence, the periods of peak vehicle and pedestrian volumes were not necessarily observed.

The video footage was reviewed in several rounds to extract the required observations for analysis. After the first two rounds, a list of vehicle arrivals, pedestrian arrivals, pedestrian departures, and PHB actuations was assembled and sorted by site and time. The beacons and pedestrian signal indications were determined for each event in this list through a series of computations using the timestamps and the known timing parameters for each PHB.

In the next rounds of video footage review, the computed beacon indications were verified, and additional detailed observations were extracted including the following:

• Vehicle position relative to the pedestrian for vehicles arriving on a steady or flashing red beacon indication.
  
  
• Driver yielding behavior during steady or flashing red beacon indication.
  
  
• Button presses by arriving pedestrians.
  
  
• Categorization of pedestrians as staged or non-staged.
  
  
• Conflict occurrences.
  
  
• Recording whether each driver arriving on steady or flashing red stopped before proceeding through the crosswalk.
  
  
• Recording whether drivers stayed stopped throughout the flashing red indication.

Additional efforts were also undertaken to record any instances of major street vehicles stopping while the beacon indication was dark as well as to identify minor road driver behaviors during a beacon actuation. The final dataset reflected over 78 h of video data and included 1,149 PHB actuations and 1,979 pedestrians who crossed the street.

DRIVER BEHAVIOR FINDINGS

Driver Behavior During Dark Indication

Selected videos were reviewed to identify each occurrence when a vehicle stopped at the crossing when the PHB was displaying a dark indication. There were several events; however, in almost all cases, it was because of congestion. There were a few cases where the driver stopped because of a bus or truck loading/unloading or because a pedestrian was in the crosswalk. Therefore, none of the drivers who stopped at the crossing when the PHB was dark appeared to be confused regarding the device.

Driver Position Relative to Pedestrian Position During Steady or Flashing Red Indications When the Driver Drove Across the Crosswalk

The position of the driver and the pedestrian for each driver that drove across the crosswalk during a steady or flashing red indication was identified. Figure 60 provides an illustration of driver and pedestrian positions for a crossing. For each cycle and for each approach, the number of cycles by approach where a given pedestrian-vehicle position combination occurred was counted. Table 62 summarizes the findings for the 1,252 cycle approaches (1,149 cycles plus 103cycles where pedestrians were crossing in both directions) for the 20 sites included in the study. The pedestrian positions were (1) edge of the street clearly indicating the desire to cross, (2) within the crosswalk on the initial approach (or first half of the crossing), or (3) within the crosswalk on the second approach (or second half of the crossing). The vehicle was either on the same approach as the pedestrian or the other approach relative to the pedestrian.

As an example, if a pedestrian was crossing a street that was oriented north and south and was on the initial approach (say the southbound approach), and if the vehicle was on the other approach, then the vehicle would be on the northbound approach. In this example, the northbound vehicle can legally enter the crosswalk during the flashing red portion of the cycle while the pedestrian is walking eastbound and crossing the southbound approach. Both Arizona and Texas State laws indicate that vehicles must yield the right-of-way to pedestrians within a crosswalk that are in the same half of the roadway as the vehicle. So drivers in cases A to F in table 62 would be considered as not yielding to the pedestrian. Both States also indicate that vehicles are to yield right-of-way if a pedestrian is approaching closely enough from the opposite side of the roadway to constitute a danger. Cases G and H in table 62 could possibly fit this situation; however, in the opinion of those reducing the data, these observations did not have the pedestrian that close to the vehicle approaching from the other approach. For all those observations, the vehicle on the other approach entered the crosswalk shortly after the steady red indication had started and the pedestrian had recently left the curb. A driver entering the intersection during the steady red indication would be in violation of the beacon indication. Cases A, C, E, G, and I reflect combinations when the driver would be in violation of the beacon.

For those cases when the driver should yield, the combination with the most yielding violations was when the pedestrian was at the edge of the street and the drivers entered the crosswalk on the steady red indication. Case A occurred for 7.7 percent of the observed cycle approaches (96 of 1,252). Most of these non-yielding (and violation) events occurred soon after the beacon changed to steady red. While it could be argued that a vehicle was not required to yield to the pedestrian in this situation since the pedestrian was not on the pavement, in the research team's opinion the pedestrians in these situation were clearly communicating the intent to cross and were not on the pavement due to safety concerns. Of course, these drivers were clearly in violation of the steady red indication. While the research team included cases A and B in the non-yielding counts, the counts are shown in table 62 so readers can draw their own conclusions. Few, but not zero as preferred, cycles had a driver entering the crosswalk when the pedestrian was on the same approach (see cases B to F). Reviewing the situations when a vehicle enters the crosswalk when the pedestrian is in the second half of their crossing (see cases E and F) revealed several occurrences when the vehicles entered the crosswalk soon after the pedestrian departed the lane.

Figure 60. Illustration. Pedestrian and driver positions when the pedestrian is on the initial approach and vehicles are present on the same approach and on the other approach.

^aPedestrian and vehicle positions are relative to where the pedestrian started the crossing (see figure 60 for an example).
^bResults reflect the percent of the 1,252 cycle approaches when the combination of pedestrian and vehicle position occurred. The total number of cycle approaches (1,252) reflect 1,149 cycles plus 103 cycles when pedestrians were crossing in both directions for the 20 sites included in the study.
^cDemonstrates whether this case reflects a non-yielding driver and/or a violation of the red indication.
^dWhile it could be argued that a vehicle is not required to yield to the pedestrian in this situation since the pedestrian was not "on the pavement," the pedestrians in these situation were clearly communicating the intent to cross and remained on the curb rather than on the pavement due to safety concerns. These drivers were in violation of the steady red indication.

Driver Yielding Behavior During Steady or Flashing Red Indications

For each pedestrian crossing when the PHB was showing steady or flashing red, the number of drivers that yielded and did not yield was determined. The driver yielding rates reflected all available pedestrian crossings regardless of whether the pedestrian was a member of the research team. A driver was considered to have not yielded to the pedestrian if the driver crossed the crosswalk markings when the PHB was in either the steady red or flashing red indications and the pedestrian was at the edge of the street clearly communicating to drivers the intent to cross or was walking on the same approach as the driver. When the crossing pedestrian was a member of the research team, the team member would place a foot on the pavement to clearly communicate the intent to cross. Using only staged pedestrian crossings would have required a much longer data collection period per site because of the large number of pedestrians crossing at the intersections. Therefore, the study included non-staged pedestrians. If it appeared that the non-staged pedestrian was not clearly communicating the intent to cross, perhaps by not standing near the edge of the sidewalk, the pedestrian crossing was not included in the study.

Counting the number of vehicles that did or did not yield to a crossing pedestrian was easier with video data when compared to gathering the data in the field. The video could be replayed to determine the exact position of a vehicle when the signal indication changed. In addition, the video allowed for greater consistency between data collectors, as an event could be reviewed by more than one person. Table 63 provides the driver yielding values for the 20 sites. Overall, driver yielding for these 20 sites averaged 96 percent. In almost all of the crossings, drivers appropriately yielded to the crossing pedestrians.

^aDriver yielding = Percent of approaching drivers who should have yielded and did so.

When reviewing the results by city, Tucson, AZ, had an average yielding rate of 97 percent while Austin, TX, had an average yielding rate of 94 percent. Affecting the average result for Austin, TX, were two sites: AU-11 and AU-27. Almost all of the non-yielding vehicles at AU-11 were northbound vehicles that crossed the crosswalk very soon after the PHB turned steady red and frequently moved at lower speeds due to high vehicle volumes present or due to an active reduced speed limit for the school zone. At AU-27, about half of the non-yielding vehicles entered the crosswalk very soon after the PHB turned to steady red (six vehicles), with the remaining non-yielding vehicles (four vehicles) entering the crosswalk before the pedestrian completely cleared that half of the roadway. Other potential differences between Austin, TX, and Tucson, AZ, that could affect yielding rates could be the length of time the PHB treatment has been used in the city (they have been in Tucson, AZ, for many more years than Austin, TX), the use of back plates (common in Tucson, AZ, not in Austin, TX), beacon face mounting locations (Tucson, AZ, typically mounted one face over the approach, one to the right of the approach, and one in the median if a raised median was present, while Austin, TX, typically mounted two faces over the approach), and the use of supplemental signs on the mast arm (Tucson, AZ, sites typically included a CROSSWALK or PEDESTRIAN CROSSWALK sign in addition to the regulatory CROSSWALK STOP ON RED (symbolic circular red) (R10-23) sign while Austin, TX, used a combination sign that provided the additional information of STOP ON FLASHING RED (symbolic flashing circular red) THEN PROCEED IF CLEAR).

There were different driver behaviors within those situations when drivers did not yield to the pedestrians. Reviewing the conditions when drivers were non-compliant revealed that several of the non-compliant drivers entered the crossing just after the PHB changed from steady yellow to steady red and the pedestrian was at the edge of the street. Given that these drivers were provided with at least 7 to 8 s of warning by way of the flashing and steady yellow indications, and based on the posted speed limits for these sites (30 to 45 mi/h), the drivers did have sufficient warning to stop upstream of the crossing but decided not to stop. In a few of the cases, the driver proceeded through the crosswalk just after the pedestrian had cleared the lane.

Driver Behavior During Flashing Red Indication

For about 20 percent of the observed PHB actuations, vehicles were not present during the flashing red indication. When a queue of vehicles was present during the flashing red indication, about half of the crossing actuations included at least one driver who did not completely stop prior to entering the crosswalk. About 5 percent of the actuations included at least one driver who stopped on the flashing red indication and remained stopped until the dark indication. This behavior was observed at about the same frequency in both cities. In some cases, these drivers might not have realized that they could proceed after stopping if their half of the crosswalk was clear of pedestrians. However, there were many cases where the stopped drivers could not proceed for one or more of the following reasons:

• The driver arrived in the last few seconds of the flashing red indication.
  
  
• The driver's half of the crosswalk was continually occupied by pedestrians, some of whom may have been slow or may have started their crossing after the start of the flashing red indication (i.e., during their flashing do not walk indication).
  
  
• Minor movement drivers were proceeding without complying with their requirements to stop.

Impact of PHB Actuation on Minor Movement Drivers

Interested practitioners have raised questions about how PHBs affect minor movements that do not pass through the beacon-controlled crosswalk. These movements may include one left-turn movement originating from the major street (LT), up to two through movements on the minor-street approaches (TH1 and TH2), a left-turn movement originating from the minor street (LT1), and a right-turn movement originating from the minor street (RT2). These movements are illustrated in figure 61. Specifically, the following questions have been asked:

• When the PHB is active, do minor movement drivers take advantage of the gaps that have been created in the major street traffic and complete their movements more easily?
  
  
• When the PHB is showing the flashing red indication, does the intersection function like a four-way stop-controlled intersection?

Figure 61. Illustration. Minor movements at a PHB-controlled crosswalk.

The research team conducted a review of the video footage to answer these questions. The review included 17 of the 20 PHB sites. Sites TU-091, AU-07, and AU-27 were excluded because they lacked the relevant movements due to site layout (midblock crossing, stop line located well downstream of the movement, or the movement is prohibited) or the movements existed but were negligible in volume. Not all minor movements were present at every site because some of the sites were three-legged intersections or the fourth leg was a driveway with minimal traffic (e.g., TU-070). The minor movements permitted at the sites are listed in table 64.

A total of 850 minor movement vehicles were observed. This set of vehicles includes only those that arrived and/or departed while the PHB was active (i.e., not displaying the dark indication). The vehicle distribution by site and movement code is provided in table 65. Note that the pairing of the LT and RT2 movements represents a significant portion of the sample size, particularly at sites TU-042, TU-072, TU-073, AU-21, and AU-22. These movements at these sites represent about 62 percent of the sample size (525 of 850 vehicles). Additional turning movements were possible but not included in the minor movement analysis. Turning vehicles that originated from the minor approaches and passed through the crosswalk are included in the other analyses documented in this chapter.

The vehicle distribution by movement code and PHB indication on vehicle arrival is provided in table 66. Arrival is defined as the time that the vehicle stopped at the stop line on its approach or the time that the vehicle crossed the stop line if it did not stop. The vehicle distribution by movement code and PHB indication on its departure is provided in table 67. Departure is defined as the time that the vehicle exited the intersection.

As shown by the total count numbers in the rightmost columns of table 66 and table 67, the turning movements were the most commonly observed movements, especially movements LT and RT2, which were complementary. The through movements were less common, partly because these movements did not exist at the three-legged intersection sites.

The total count numbers in the bottom rows of table 66 and table 67 generally reflect the proportion of time that the PHB was active. The number of vehicles arriving or departing on the dark indication is small because vehicles that both arrived and departed on the dark indication were excluded from this analysis.

The distribution of minor movement vehicles by PHB indication upon arrival and PHB indication upon departure is shown in table 68. As shown, most vehicles departed on either steady red or flashing red regardless of when they arrived. However, a small number of vehicles experienced notable delay because they arrived while the PHB was active but could not depart until the next dark indication. Specifically, nine vehicles arrived during flashing red but did not depart until the steady yellow or steady red indication during the next PHB actuation. These long delays occurred during periods when the major street volume was sufficiently high enough that no acceptable turning or crossing gaps were available during the dark indication, and the major street drivers were aggressive during the flashing red indication such that they either did not stop consistently or stopped briefly and proceeded without waiting for other drivers' movements to proceed.

To examine stop compliance, drivers were classified as violators under the following conditions:

• The driver was making the LT movement, arrived during a steady or flashing red indication, and did not stop. At all sites, the stop line was marked across all through and left-turn lanes, such that these drivers were required to stop even though they do not pass through the crosswalk.
  
  
• The driver was making the LT1, TH1, TH2, or RT2 movement and did not stop. In other words, the driver failed to stop at the STOP sign.

Each driver in the database was thus classified as a violator or non-violator, and violation rates were computed for each site based on the amount of video footage that was collected at the site (i.e., violations per hour). These findings are shown in table 69. The percentage of drivers not stopping varied widely, from 30.8 to 89.9 percent. Many of these percentages were computed based on the small number of minor movement vehicles that were observed while the PHBs were active. In terms of rate, violations at most sites were less common than five violations/h (or one violation every 12 min).

The following seven sites were found to have violation rates in excess of five violations/h: TU-042, TU-059, TU-073, AU-04, AU-21, AU-22, and AU-24. A more focused examination of these sites was conducted by computing violation rates for each movement. The results of this examination are shown in table 70. Similar computations were performed for all movements at all sites, but none of the movements omitted from table 70 had violation rates higher than four violations/h.

All of the movements listed in table 70 were located close to major traffic generators such as schools or supermarkets. In fact, the majority of the violations observed at site TU-042 occurred in the morning period before classes began for the day. Violations were also frequent at site AU-22; however, it should be noted that the LT movement at this site occurred from a TWLTL, and the stop line for the PHB was marked in the adjacent through lanes but not in the TWLTL, suggesting that the stopping requirement might not have been intended for this movement. This marking practice differed from the practice at other Austin, TX, sites, where stop lines were extended through all approach lanes. Sites AU-21 and AU-22 were similar in that their LT and RT2 movements went into or out of supermarkets, but the LT movement at site AU-22 did not have a high violation rate. The major street at this site was a four-lane undivided city arterial, so left-turning drivers did not have a turn bay and were often blocked from completing the LT movement while the PHB was active.

Based on the preceding information, the following observations could be made:

• Minor movement drivers took advantage of gaps that were created in the major street traffic while the PHBs were active.
  
  
• PHB sites did not necessarily operate like four-way stops while the flashing red indication was shown. At sites near major traffic generators, movements entering or exiting the traffic generator tended to dominate the operation of the intersection, and stop compliance for these movements tended to be low.

These trends may occur at any type of unsignalized pedestrian crossing treatment while pedestrians are present. They are likely not unique to PHB-controlled crossings.

PEDESTRIAN BEHAVIORS FINDINGS

Pedestrian Departures by Indication

Of the 1,979 pedestrians crossing the street, 290 were research team members who always crossed when the beacons showed steady or flashing red to the motorists. Most of the remaining 1,689 general public pedestrians departed when the beacon showed a steady red indication to the drivers. As shown in table 71, 80 percent departed during the steady red or flashing red indications. Approximately 13 percent of the pedestrians departed while the PHB was still in the vehicle clearance intervals.

^aThe PHB was located within the coordinated corridor where the timing of when the PHB was active was influenced by the nearby coordinated signals.

For the pedestrian crossings observed that did not include the research team member crossings, only 124 pedestrians (7 percent) left during the dark indication. For the majority of these pedestrians, the roadway volume was such that the pedestrian was able to find sufficient gaps to cross. The volume per minute per lane was less than fourvehicles/min/lane for the majority of these crossings. Figure 62 shows the cumulative distributions of the 1 min/lane volume for those pedestrians that departed during the dark indication (i.e., blue dashed line) and those that departed during an active indication (i.e., red solid line). Pedestrians were more likely to wait for the PHB to be active before starting to cross at the higher roadway volumes, as shown by the location of the red solid line to the right of the blue dashed line. For example, the cumulative distributions reached the value of 80 percent at volumes of about six vehicles/min/lane for the blue dashed line and about eight vehicles/min/lane for the red solid line. This trend shows that only 20 percent of non-compliant pedestrians were still willing to cross if the roadway volume exceeded six vehicles/min/lane (i.e., 1,440 vehicles/h at a four-lane site). Less than 5 percent of non-compliant pedestrians crossed when the roadway volume exceeded 10 vehicles/min/lane. Conversely, roughly half of the compliant pedestrians crossed at roadway volumes exceeding six vehicles/min/lane.

Figure 62. Graph. Volume cumulative distribution when pedestrian started the crossing.

Two of the PHB sites were located within coordinated corridors where the timing of PHB activation was influenced by the nearby coordinated signals. The other 18 sites were at "hot button" sites, where the PHB sequence starts following the push of the pushbutton. About one-third of the pedestrians (36 of 124) who departed during the dark indication were at the sites where the PHB is coordinated with nearby signals. While a large value, there were other sites with more pedestrians departing during the dark indication, both in terms of number of pedestrians and proportion of pedestrians observed at the site. The percentage of pedestrians departing during the dark indication was 13 percent at the coordinated sites and 6 percent at the hot button sites. Departures on dark were much less frequent at the coordinated site that had pedestrian pushbuttons with red lights that illuminate when the button is pressed. The coordinated site with the red-lighted buttons had 10 percent of pedestrians departing on dark, while the coordinated site with the non-lighted buttons had 20 percent of pedestrians departing on dark.

Pedestrian Actuation of the PHB

Of the 1,979 pedestrians crossing the street, 290 were research team members who crossed using a staged pedestrian protocol and always activated the PHB. The remaining 1,689 general public pedestrians were coded by whether they pushed the pedestrian pushbutton or did not push the pushbutton subdivided by whether the PHB was already active or not active when they arrived to the crossing. Table 72 shows the number of pedestrians by action. Overall, most pedestrians (average of 91 percent) who could have activated the PHB did. As can be seen in table 72, there were some sites with lower percent actuations (e.g., AU-04) and other sites where every pedestrian crossed with an activated PHB (e.g., TU-03, TU-04, and AU-11).

^aPercent = Percentage reflecting the ratio of the number of pedestrians that pushed the button to the number of pedestrians that if they would have pushed the button would have triggered a change in the device (i.e., those that pushed and those that did not when the device was dark). In other words, this column does not include those pedestrians who arrived when the PHB was active.
^bThe PHB was located within coordinated corridor where the timing of when the PHB was active was influenced by the nearby coordinated signals.

A plot of the percentage of pedestrians who activated the PHB when arriving to the crossing when the PHB was not already active is shown by posted speed limit in figure 63, by crossing distance in figure 64, and by hourly volume in figure 65 . A review of these plots shows trends for the highest values. A high number of pedestrians (93 percent) activated the device on the 45-mi/h posted speed limit road. For the 40-mi/h or less roads, a large range of actuation was observed—between 75 and 100 percent. The percentage of pedestrians pushing the button was always greater than 83 percent for the longer crossing distances (i.e., longer than 110 ft).

The 1-min volume count nearest to the arrival time of the pedestrian was determined. The number of pedestrians by their action was summed for each 1-min count value for all 20 sites. The 1-min counts with less than 20 pedestrians crossing were omitted for the plot shown in figure 65. The 1-min count was adjusted to an hourly equivalent value by multiplying the 1-min count by 60. When the hourly volume for both approaches was 1,500 vehicles/h or more, the percent of pedestrians activating the PHB was always 92 percent or more.

Figure 63. Graph. Percentage of pedestrians pushing the button, by posted speed limit.

Figure 64. Graph. Percentage of pedestrians pushing the button, by crossing distance.

Figure 65. Graph. Percentage of pedestrians pushing the button, by 1-min volume counts adjusted to hourly counts.

Some of the pedestrians who activated the PHB departed prior to the walk indication. The beacon indication when the pedestrian departed is shown in table 73. Most of the pedestrians who departed after pushing the pushbutton either left during the steady red (82 percent) or the flashing red (2 percent). About 1 percent of those who activated the device left while the device was still dark, with most occurring at the two locations where the PHB was within a coordinated corridor (sites TU-072 and AU-07).

^aThe PHB was located within coordinated corridor where the timing of when the PHB was active was influenced by the nearby coordinated signals.

CONFLICTS FINDINGS

All occurrences of pedestrian/vehicle conflicts and erratic maneuvers were noted when observed in the video footage. A conflict is defined based on the following criteria:

• The vehicle suddenly changes path, speed, or both.
  
  
• The vehicle's brake lights illuminate unexpectedly during a turning maneuver.
  
  
• The pedestrian exhibits one or more of the following behaviors:
  
  
  • Slows down due to vehicle presence.
    
    
  • Stops and waits for the vehicle to pass through the crosswalk.
    
    
  • Steps sideways to avoid a vehicle.
    
    
  • Runs or speeds up to avoid a vehicle.
    
    
  • Turns back to origin curb to avoid a vehicle.
    
    
  • Makes another sudden change in path or speed in response to an approaching vehicle.

In the 78 h of video footage, 54 conflicts were observed. The distribution of the conflicts categorized by vehicle maneuver was 38 for through vehicles, 10 for left-turning vehicles, and 6 for right-turning vehicles. Distribution by vehicle maneuver was analyzed because of concerns about turning traffic operations while the PHB was serving pedestrians. While major street through vehicles were stopped by the PHB, turning drivers originating from the minor approaches could see significant gaps form on the major street and may have taken this opportunity to turn, possibly leading to conflicts with pedestrians in the crosswalk.

The conflicts were tabulated by vehicular beacon indication and vehicle maneuver, as shown in table 74. Beacon indication can be used to classify pedestrians as compliant (i.e., they departed the corner on a steady red indication while the walk indication was displayed) or non-compliant (i.e., they departed on one of the other beacon indications while the flashing or steady do not walk indication was displayed). Slightly less than half of the observed conflicts occurred during the dark beacon indication and involved a through vehicle. These conflicts usually involved pedestrians who either crossed without pushing the button or pushed the button but did not wait for the walk indication and then paused in the raised curb median while crossing. The latter case occurred most frequently at site TU-072 (14 conflicts), which had 6 lanes and operated in coordinated mode.

^aIndicates pedestrians who were not in compliance with signal indication.

A notable number of conflicts involving left-turning vehicles was observed at site AU-21. This site was located near a well-patronized supermarket and a bus stop, with a driveway located about 45 ft away from the crosswalk. At that distance, drivers making left turns out of the supermarket parking lot would be able to turn onto the major street but would still be oriented diagonally when encountering the crosswalk. Many of these drivers encroached on the crosswalk while attempting to complete their turning maneuvers, leading to conflicts if pedestrians were present. A similar geometric layout was present at site AU-22, but this site did not exhibit conflicts because the driveway was located farther away from the crosswalk (about 60 ft) and afforded left-turning drivers more room to complete their turning maneuvers and wait for pedestrians to clear.

To account for exposure differences across the sites, conflict rates were computed using three conflict rate measures, as provided in table 75. The measure of conflicts per pedestrian-vehicle accounted for the influence of both pedestrian and vehicle exposure. For all three rate measures, the conflict rate was found to be higher for non-compliant pedestrians than for compliant pedestrians. These rates were interpreted in terms of total exposure (pedestrians or vehicles or pedestrian and vehicles). That is, at a site that experiences 100 pedestrians crossing per hour, it is expected that 2.73 conflicts would be observed per hour, of which 1.06 would involve compliant pedestrians and 1.67 would involve non-compliant pedestrians.

Notable conflict rates were observed at site TU-072 because of the high number of non-compliant pedestrians. This site operated in coordinated mode with buttons that did not provide audible or visual cues indicating that the button press had been registered. As a result, many pedestrians pushed the button but may have believed that the PHB was malfunctioning or may have grown impatient while waiting for service and crossed the major street before the walk indication was displayed. Similar behavior was not observed at site AU-07, which operated in coordinated mode but with buttons that had small red lights that illuminated when the button was pressed.

Notable conflict rates for both compliant and non-compliant pedestrians were also observed at several sites where the PHBs were located near supermarkets and multiple bus stops. At these sites, many bus riders walked through the supermarket parking lots or ran across the major street while transferring between bus lines. The presence of bus stops near an access point with significant turning vehicle volumes tended to result in higher conflict rates.