Accuracy and speed in identifying WALK indication for a given crossing


Access Board Draft PROWAG and the APS Guide emphasize the importance of installation location for APS. Installation location is critical for accurate and fast identification of which crosswalk has the WALK signal. Research on WALK signal location and tone, described in this section, was the basis for that recommendation.

APS must provide unambiguous information regarding which crosswalk has the WALK indication. A pedestrian who mistakes the signal for one crossing from a corner, for the signal for the other crossing, is at risk of making a crossing when vehicular traffic has the right-of-way and pedestrian crossing is not permitted. A two-tone system such as cuckoo/cheep has long been assumed to provide unambiguous information, however San Diego surveys and Uslan et al. noted problems with this system. ACB and AER surveys confirmed that there were numerous problems with it.

Speed in identifying the WALK signal for the desired crossing is related to delay in starting to cross. The most desirable signal from this perspective is one that is quickly identified, enabling users to initiate crossings promptly, and to finish crossing before the onset of perpendicular traffic.

Figure C-1. Positions and headings of loudspeakers and pedestrians at each corner in NEI research. (Figure from Ashmead, et al., 2004) Note: The question to the pedestrian always was, "Which has the WALK signal, the crosswalk straight in front of you or the one to your right?" Corner 1: loudspeakers near curb, on outside of crosswalk line. Corner 2: loudspeakers near back edge of sidewalk, on outside of crosswalk line. Corner 3: loudspeakers near curb but facing across pedestrian's position. Corner 4: loudspeakers near back edge of sidewalk mounted on the same pole. The figure is not drawn to scale.

Figure C-2. Optimal location of pushbutton-integrated APS.

Figure C-3. Locations of pushbutton-integrated APS and associated WALK signals in NCHRP 3-62 research

Results of research

In NEI research, Ashmead, Wall, Bentzen and Barlow (2004) investigated the effects on accuracy and speed of identifying the correct crosswalk, of location of APS speakers located in different positions relative to the crosswalk (see figure C-1 below for APS positioning used in research), using typical placements seen in the U.S.

Under the most typical signal mode, simultaneous presentation from both ends of the crosswalk, the most accurate performance occurred when signals were placed close to the curb line, near the side of the crosswalk that was furthest from the center of the intersection (see corner 1 in figure C-1). The pedestrian could easily tell which of the two loudspeakers at the corner was active because each loudspeaker was close to the position of the pedestrian waiting to cross at the associated crosswalk. Note that this was true despite the fact that both signals on the corner had the same sound. Other arrangements of loudspeakers resulted in somewhat worse performance. Accuracy at corner 3 was poor, with participants answering correctly on only 25% of trials in the simultaneous presentation condition and 50% of trials in the alternating presentation condition.

There was no evidence that response time differed across the four corners, that is, for different speaker arrangements. This suggests that the inaccurate judgments made from corner 3 about which crosswalk had the WALK signal did not reflect uncertainty, but rather were mistakes of which the participants were largely unaware.

The above graphic illustrates the recommended placement of APS devices in relation to the crosswalk provided in the APS Synthesis in 2003. This arrangement was based on the NEI research described above. However, the Panel for the NCHRP Project 3-62 asked for additional research to determine if use of two poles at each corner was necessary, or if installation of two devices on one pole could result in accurate and fast identification of which crosswalk had the WALK indication.

Both the location of pushbutton-integrated APS, and various associated signals were investigated under NCHRP 3-62 at an intersection in Portland, Oregon. (See Figure C-3 for the locations of APS on each corner, and the sounds associated with each.). Ninety participants who were blind, who had low vision, or who had cognitive disabilities made all eight approaches to a 4-way intersection, were asked to push the button for to cross the street in front of them, and then to indicate when the WALK signal came on for the street in front of them. (Participants did not cross the street except as they were guided from one corner to the next). (See 3-62 Final Report and Scott, Myers, Barlow and Bentzen, 2006.)

Results indicated that where pushbutton-integrated APS were mounted on separate poles, near the crosswalk line furthest from the center of the intersection, approximately 3 feet from the curb line, and approximately 10 feet apart, accuracy in judging when the correct crosswalk had the WALK signal was significantly better than when APS were located according to other criteria (see corner C, Figure C-3). On Corner C, participants in the totally blind and legally blind subgroups indicated that the street in front of them has the WALK signal, when in fact the WALK signal was to cross the street beside them, on only 7.55% of trials. For all other tested arrangements, the same participants made this type of error on at least 26.9% of trials. Responses to the onset of the correct WALK signal were also significantly faster at corner C.

For the two subgroups with the least vision, use of two different tones, when APS speakers were located on the same pole, resulted in errors in determining which street had the WALK interval on 50% of trials. Even when APS were located on two separated poles, accuracy in identifying the onset of the correct WALK signal was significantly greater when both APS on the same corner used the same tone (rapid tick) than when the two APS used two different tones (cuckoo and rapid tick). Speech WALK messages, when two APS were located on the same pole, were also evaluated in this research. Speech messages resulted in a much lower error rate than two tones, 19% vs 50%, however locating the APS on two separated poles and using the same tone on each for the WALK indication resulted in an error rate of less than 4%. Locating the APS speakers close to the crosswalk being signaled resulted in much better accuracy in identifying which street had the WALK interval than variation in WALK indications. The research recommended use of the rapid tick WALK indication because it produced the fastest and most accurate responses regarding which crosswalk has the WALK indication. This recommendation was in conjunction with recommendations for specific locations for the APS.

Participants who had enough vision to see the pedestrian signal (identified in the study as low vision or cognitively impaired) reported that they used the visual signal. However, the results for these groups are in the same direction as the results for the two subgroups with the least vision. It thus appears that the participants who could see the visual pedestrian signals might nonetheless have been influenced by the APS pole arrangement and signal sound.

In research comparing devices having various features, conducted in Tucson and Charlotte under NCHRP 3-62, (see NCHRP 3-62 Final Report; Barlow et al., 2005; Bentzen et al., in press), there were three types of WALK indications used in the devices, speech messages, birdcalls (cuckoo-cheep), and the rapidly repeating tone indication (rapid tick). Once pedestrians understood the crossing signal, the rapid tick provided the best cue in terms of starting to cross quickly in both cities. The faster response to the rapid tick signal confirms results of previous research on pushbutton location and nature of WALK signal, in which responses were faster to the tick than to two different tones or to speech messages.

[back to top]