Guidelines for the Application of Crossing Solutions at Roundabouts and Channelized Turn Lanes for Pedestrians with Vision Disabilities (2016)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

NCHRP 3-78b: Final Project Report April 2016 41 4 FIELD STUDY RESULTS This chapter presents a summary of the field studies for the eight roundabout and 12 channelized turn lane sites included this research. The first part of the chapter presents the actual field study results in the form of simple lists accompanied by photos. The second part of the chapter provides a narrative of observations for the various sites. Additional detailed results are provided in Appendix E. 4.1 Summary of Field Studies This section presents a summary of the field study results, starting with the channelized turn lane sites, and followed by the roundabout sites. The results cover the results of the crossing indicator study with blind participants (producing measures of intervention rate and delay), yielding study (producing measure of yielding rate), and vehicle free-flow speed study (producing the average speed of vehicles at the crosswalk). Results for the wayfinding studies are discussed in Section 4.2 and Appendix A. Results of the site photo logs are presented in Appendix E. Additional field study details are provided in Appendix F. 4.1.1 Channelized Turn Lane Sites A total of 12 channelized sites were studied in this research, with sites located in Tucson, AZ (4 sites), Boulder, CO (6 sites), Greenbelt, MD (1 site), and Cary, NC (1 site). The results are presented below in sequence. 4.1.1.1 Tucson, AZ Four channelized turn lanes were studied in Tucson, AZ as listed below. 1. Wilmont Road at Speedway Boulevard – Northwest quadrant of intersection, urban location, deceleration lane, no acceleration lane, stop sign at downstream merge point a. Intervention Rate: 2.2% b. Average Delay: 6.3 seconds c. Yielding Rate: 88.6% d. Average Speed (200’): G – 33 mph; R – 28 mph e. Average Speed (CW): n/a (stop sign) Figure 4-1: Wilmont Road at Speedway Boulevard, NW Quadrant, Tucson, AZ

NCHRP 3-78b: Final Project Report April 2016 42 2. Sabino Canyon Road at Tanque Verde Road – Northeast quadrant of intersection, urban location, deceleration and acceleration lane, no additional treatments a. Intervention Rate: 0.0% b. Average Delay: 4.2 seconds c. Yielding Rate: 46.6% d. Average Speed (200’): G – 30 mph; R – 28 mph e. Average Speed (CW): G – 22 mph; R – 19 mph Figure 4-2: Sabino Canyon Road at Tanque Verde Road, NE Quadrant, Tucson, AZ 3. Grant Road at Oracle Road – Northeast quadrant of intersection, urban location, deceleration lane, no acceleration lane, raised crosswalk a. Intervention Rate: 0.0% b. Average Delay: 3.0 seconds c. Yielding Rate: 67.9% d. Average Speed (200’): G – 31 mph; R – 30 mph e. Average Speed (CW): G – 21 mph; R – 19 mph Figure 4-3: Grant Road at Oracle Road, NE Quadrant, Tucson, AZ 4. Grant Road at Oracle Road – Southwest quadrant of intersection, urban location, deceleration lane, no acceleration lane, raised crosswalk a. Intervention Rate: 6.7% b. Average Delay: 3.7 seconds c. Yielding Rate: 49.5% d. Average Speed (200’): G – 32 mph; R – 29 mph

NCHRP 3-78b: Final Project Report April 2016 43 e. Average Speed (CW): G – 20 mph; R – 17 mph Figure 4-4: Grant Road at Oracle Road, SW Quadrant, Tucson, AZ 4.1.1.2 Boulder, CO Six channelized turn lanes were studied in Boulder, CO as listed below. 1. 28th Street at Pearl Street – Northeast quadrant of intersection, urban location, deceleration lane, no acceleration lane, no additional treatments a. Intervention Rate: 8.5% b. Average Delay: 12.2 seconds c. Yielding Rate: 66.0% d. Average Speed (200’): G – 25 mph; R – 21 mph e. Average Speed (CW): G – 14 mph; R – 13 mph Figure 4-5: 28th Street at Pearl Street, NE Quadrant, Boulder, CO 2. 28th Street at Pearl Street – Northwest quadrant of intersection, urban location, deceleration lane, no acceleration lane, raised crosswalk a. Intervention Rate: 1.7% b. Average Delay: 15.7 seconds c. Yielding Rate: 57.8% d. Average Speed (200’): G – 26 mph; R – 20 mph e. Average Speed (CW): G – 15 mph; R – 13 mph

NCHRP 3-78b: Final Project Report April 2016 44 Figure 4-6: 28th Street at Pearl Street, NW Quadrant, Boulder, CO 3. 28th Street at Canyon Boulevard – Southwest quadrant of intersection, urban location, deceleration lane, no acceleration lane, raised crosswalk a. Intervention Rate: 6.7% b. Average Delay: 23.9 seconds c. Yielding Rate: 40.3% d. Average Speed (200’): G – 29 mph; R – 27 mph e. Average Speed (CW): G – 15 mph; R – 14 mph Figure 4-7: 28th Street at Canyon Boulevard, SW Quadrant, Boulder, CO 4. Foothills Parkway at Arapahoe Avenue – Southwest quadrant of intersection, suburban location, deceleration lane, acceleration lane, raised crosswalk a. Intervention Rate: 0.0% b. Average Delay: 14.6 seconds c. Yielding Rate: 32.4% d. Average Speed (200’): G – 33 mph; R – 35 mph e. Average Speed (CW): G – 21 mph; R – 22 mph

NCHRP 3-78b: Final Project Report April 2016 45 Figure 4-8: Foothills Parkway at Arapahoe Avenue, SW Quadrant, Boulder, CO 5. Foothills Parkway at Baseline Drive – Southwest quadrant of intersection, suburban location, deceleration lane, acceleration lane, sound strip treatment a. Intervention Rate: 0.0% b. Average Delay: 9.8 seconds c. Yielding Rate: 30.5% d. Average Speed (200’): G – 28 mph; R – 29 mph e. Average Speed (CW): G – 20 mph; R – 18 mph Figure 4-9: Foothills Parkway at Baseline Drive, SW Quadrant, Boulder, CO 6. Foothills Parkway at Baseline Drive – Northeast quadrant of intersection, suburban location, deceleration lane, acceleration lane, raised crosswalk a. Intervention Rate: 0.0% b. Average Delay: 13.0 seconds c. Yielding Rate: 36.1% d. Average Speed (200’): G – 33 mph; R – 31 mph e. Average Speed (CW): G – 25 mph; R – 22 mph

NCHRP 3-78b: Final Project Report April 2016 46 Figure 4-10: Foothills Parkway at Baseline Drive, NE Quadrant, Boulder, CO 4.1.1.3 Greenbelt, MD and Cary, NC Finally, one channelized turn lane each were studied in Greenbelt, MD and Cary, NC as listed below. 1. Kenilworth Avenue at East West Highway - Northwest quadrant of the intersection, deceleration lane, no acceleration lane, no other treatments installed a. Intervention Rate: 10.4% b. Average Delay: 20.1 seconds c. Yielding Rate: 23.2% d. Average Speed (200’): G – 28 mph; R – 29 mph e. Average Speed (CW): G – 17 mph; R – 15 mph Figure 4-11: Kenilworth Avenue at East West Highway, NW Quadrant, Greenbelt, MD 2. Kildaire Farm Road at Tryon Road – Southwest quadrant of the intersection, deceleration lane, no acceleration lane, no other treatments installed a. Intervention Rate: 3.3% b. Average Delay: 16.0 seconds c. Yielding Rate: 46.8% d. Average Speed (200’): G – 38 mph; R – 35 mph e. Average Speed (CW): G – 15 mph; R – 13 mph

NCHRP 3-78b: Final Project Report April 2016 47 Figure 4-12: Kildaire Farm Road at Tryon Road, SW Quadrant, Cary, NC 4.1.2 Roundabout Sites A total of 8 roundabout approaches were studied in this research, with each location featuring an entry and an exit leg for a total of 16 data points. The sites were located in Hilliard, OH (2 entry legs, 2 exit legs), Oakland County, MI (2 entry legs, 2 exit legs), Greenbelt, MD (1 entry leg, 1 exit leg), Ann Arbor, MI (2 entry legs, 2 exit legs) and Cary, NC (1 entry leg, 1 exit leg). The results are presented below in sequence with photos. 4.1.2.1 Hilliard, OH 1. Main Street at Cemetery Road – East and west approaches, two-lane entries and exits, west approach features a “standard” crosswalk location, east approach features an offset or “zig- zag” configuration, both approaches have in-road yield to pedestrian signs a. Intervention Rate: • East Entry: 1.7% • East Exit: 6.7% • West Entry: 1.9% • West Exit: 14.0% b. Average Delay: • East Entry: 21.7 seconds • East Exit: 17.4 seconds • West Entry: 14.8 seconds • West Exit: 17.9 seconds c. Yielding Rate: • East Entry: 58.9% • East Exit: 23.4% • West Entry: 63.6% • West Exit: 21.3% d. Average Speed (CW): • East Entry: 17 mph • East Exit: 26 mph • West Entry: 16 mph • West Exit: 21 mph

NCHRP 3-78b: Final Project Report April 2016 48 Figure 4-13: Main Street at Cemetery Road, East Approach, Hilliard, OH 4.1.2.2 Oakland County, MI 1. Maple Road at Farmington Road – East and south approaches, RRFB on all approaches, raised crosswalk on tested approaches only, three-lane east leg, two-lane north and south legs a. Intervention Rate: • East Entry: 0.0% • East Exit: 0.0% • South Entry: 0.0% • South Exit: 6.0% b. Average Delay: • East Entry: 9.4 seconds • East Exit: 10.9 seconds • South Entry: 9.3 seconds • South Exit: 8.2 seconds c. Yielding Rate: • East Entry: 90.8% • East Exit: 54.1% • South Entry: 65.2% • South Exit: 65.4% d. Average Speed (CW): • East Entry: 13 mph • East Exit: 15 mph • South Entry: 13 mph • South Exit: 15 mph

NCHRP 3-78b: Final Project Report April 2016 49 Figure 4-14: Maple Road at Farmington Road, East Approach Exit, Oakland County, MI Figure 4-15: Maple Road at Farmington Road, South Approach Entry, Oakland County, MI 4.1.2.3 Greenbelt, MD 1. Greenbelt Metro Drive at Cherrywood Lane – West entry and west exit approaches, raised crosswalks, one-lane entry, two-lane exit a. Intervention Rate: • West Entry: 2.1% • West Exit: 4.0% b. Average Delay: • West Entry: 24.0 seconds • West Exit: 26.2 seconds c. Yielding Rate: • West Entry: 41.9% • West Exit: 13.7% d. Average Speed (CW): • West Entry: 17 mph • West Exit: 17 mph

NCHRP 3-78b: Final Project Report April 2016 50 Figure 4-16: Greenbelt Metro Drive at Cherrywood Lane, West Entry and Exit Approaches, Greenbelt, MD 4.1.2.4 Ann Arbor, MI 1. Ellsworth Road at State Road – West entry and west exit approaches, “zig-zag” crosswalk, rumble strips a. Intervention Rate: • West Entry: 0.0% • West Exit: 3.1% b. Average Delay: • West Entry: 7.9 seconds • West Exit: 9.9 seconds c. Yielding Rate: • West Entry: 78.3% • West Exit: 8.0% d. Average Speed (CW): • West Entry: 18 mph • West Exit: 27 mph Figure 4-17: Ellsworth Road at State Road, West Entry and Exit Approaches, Ann Arbor, MI 2. Nixon Road at Huron Parkway – South entry and south exit approaches, single-lane roundabout, rumble strips a. Intervention Rate: • West Entry: 0.0%

NCHRP 3-78b: Final Project Report April 2016 51 • West Exit: 0.0% b. Average Delay: • West Entry: 5.8 seconds • West Exit: 8.4 seconds c. Yielding Rate: • West Entry: 79.4% • West Exit: 45.5% d. Average Speed (CW): • West Entry: 15 mph • West Exit: 16 mph Figure 4-18: Nixon Road at Huron Parkway, South Exit and Entry Approaches, Ann Arbor, MI 4.1.2.5 Cary, NC 1. Old Apex Road at West Chatham Street – West entry and west exit approaches, three-legged, single-lane roundabout a. Intervention Rate: • West Entry: 1.7% • West Exit: 3.3% b. Average Delay: • West Entry: 11.4 seconds • West Exit: 11.7 seconds c. Yielding Rate: • West Entry: 61.4% • West Exit: 32.2% d. Average Speed (CW): • West Entry: 18 mph • West Exit: 21 mph

NCHRP 3-78b: Final Project Report April 2016 52 Figure 4-19: Old Apex Road at West Chatham Street, West Entry and Exit Approaches, Cary, NC 4.2 Field Observations and Descriptive Data on Wayfinding As discussed earlier, blind participants were asked to find the crosswalk and to cross at the roundabout and CTL sites after completing the indicator trials. The wayfinding trials were conducted at a subset of intersections for the indicator trials. Before the indicator trials, participants were given a general orientation to the roundabout or CTL, which focused on traffic movement characteristics. Details of the crosswalks or islands were not described or explored as part of the orientation. At most sites, the participant had walked up to the crosswalk numerous times during the indicator trials, so some participants were familiar with one approach. However, they had not approached from both directions and had only crossed to the island once or twice to get an idea of the width of the lane they were judging. Each participant completed two approaches at each location, usually one from each of the two possible approach directions, with the crosswalk ahead on the right on one approach and on the left on the other approach. There were some locations where the sidewalk did not continue past the crosswalk, so both approaches were made from the same direction. The actions of the participants were observed and recorded, as detailed in Appendix A, and observations were recorded by researchers, who were certified orientation and mobility specialists (COMS), about the common problems observed and the features that seemed to affect wayfinding behavior. This observational information was generally supported the descriptive statistics reported in Appendix A. 4.2.1 Wayfinding Features – Summary of Observational Data Summaries of the observations made by O&M specialists during the wayfinding trials are organized below by the types of errors observed during the data collection. The comments and impressions of blind participants were also noted in discussion with researchers after each set of wayfinding trials, and are summarized here and in Appendix A. The description of the observations is organized by the wayfinding tasks of interest: 1) Determining the Crossing Location, including detecting the street, 2) Aligning to Cross and Establishing a Correct Heading, and 3) Maintaining Correct Heading While Crossing (e.g., staying within the crosswalk). An additional section discusses features of intersections that may affect wayfinding for persons who are blind, such as the pushbutton location of pedestrian signals. Tasks 1-3 were completed from the curb to the island, while only the first and second tasks were completed after participants had crossed to the island. Some observations relate mainly to the island wayfinding features

NCHRP 3-78b: Final Project Report April 2016 53 Note that islands that are delineated only by pavement markings are not recognized at all by pedestrians who are blind. 4.2.2 Determining the Crossing Location Typical strategies at signalized and stop-controlled intersections for locating a curb ramp or crossing location leading to a crosswalk include continuing to the curb in the direction of travel while approaching the intersection, and identifying traffic stopped at the stop line on the perpendicular street as a cue to the crosswalk location. At roundabouts this strategy can prove difficult to implement. Because of the curving nature of the sidewalk, individuals may find it difficult to recognize the desired crossing point. To locate the crosswalk at a roundabout or CTL, participants often had to follow (tactually, using their cane) a landscape edge to their side, or the curb itself, using various cues to locate the crossing location. These cues may include changes in the landscaping, curb ramps, or the presence of detectable warning surfaces. It appeared that most participants were unaware of the need to walk on the side of the sidewalk closest to the street to find the crossing location. Also, because most of the cues to the location of the crossing point do not extend for the full width of the sidewalk, they will not likely be detected by pedestrians who are blind, if they are walking on the sidewalk and away from the street. When prompted that they had passed the crosswalk, many participants began to follow or “trail” the edge of the landscaping near the curb, or follow the curb itself. The second aspect of locating the crosswalk location is detecting the street and stopping before stepping into the street. Recognizing the curb or the slope of the curb ramp, the detectable warning surface, and the gutter along the edge of the street, are typically the key cues used to recognize the edge of the street, either at the curb edge or when crossing from the islands. Whether considering the data in Appendix A, Figures 14-1 through Figure 14-6, or the data for each intersection location as shown in Tables 14-4 through Table 14-7, it is clear that determining the crossing location can be a significant problem for many blind pedestrians. For example, Table 14-4 and Table 14-6 show that large percentages of participants missed the crosswalk at most of the roundabout sites. This was true both for the crosswalk leading to an island and for crosswalks leading from the island to the curb. However, there were a few exceptions to this general finding at some roundabout locations. Also, like these exceptions at roundabouts, Table 14-5 indicates that, for the CTL sites, most of the crossing locations on trials to the island were successfully located by the participants. It is unclear what factors accounts for the wide variations across sites and individuals. It appears from the research conducted in this project that the design of an accessible roundabout or CTL for persons who are blind requires additional study in order to better understand why determining the crossing location is a substantial challenge when traveling without vision, and to determine what travel strategies and environmental modifications aid individuals in successful completion of this key wayfinding task. 4.2.2.1 Features that appeared to aid in determining the crossing location (locating the crosswalk) a) Grass or other landscape strip between sidewalk and curb When there was a landscape strip between the sidewalk and curb, blind pedestrians usually did not walk on it, and instead stayed on the sidewalk while looking to the side with their cane for the crosswalk. Participants reported that they were more comfortable when looking for the crosswalk when they were separated by landscaping from traffic moving in the roadway. Not all participants used the strategy of following, or trailing the landscape strip on their side closest to the roadway with their cane when looking for the opening to the crosswalk at first. Most participants adopted that strategy (if there was a landscape strip) after they were told that they had passed the crosswalk.

NCHRP 3-78b: Final Project Report April 2016 54 Figure 4-20: Grass between sidewalk and curb along edge of circulatory roadway between the crosswalks This figure shows a photo of a roundabout with an approximately 4-foot-wide grass strip between the sidewalk and the roadway that follows the curvature of the road between the crosswalks at a roundabout. b) Grass or gravel outside the crosswalk area, particularly on islands. If there was grass or a gravel type surface that felt different under foot than pavement, blind participants usually did not continue walking on it, but looked with their cane of foot for a sidewalk-type surface on which to travel. Less disorientation and confusion was noted on islands with landscaped or graveled areas. Figure 4-21: Gravel landscape strip provides detectable edge of sidewalk for person who is blind and separation between the sidewalk and CTL travel lane This figure shows a photo of an approximately 3-foot-wide gravel strip between the sidewalk and the roadway that follows the curvature of the road to the crosswalk at a CTL.

NCHRP 3-78b: Final Project Report April 2016 55 Figure 4-22: Gravel or grass outside the crosswalk and walkway area Photo of a curb ramp leading to a crosswalk, with detectable warning surface on the ramp, returned curb on the ramp and gravel landscaping outside the sidewalk area. 4.2.2.2 Features that didn’t seem to provide adequate information to pedestrians who were blind in locating the crosswalk a) Paved or hardscape surfaces For pedestrians who were blind, paved or hardscape surfaces, even with those with had a relatively rough texture, were not observed to provide a reliable cue that the area was not the sidewalk area. At one roundabout location, cobblestone pavers were used for separation between the roadway and sidewalk, as shown in the photo below, but blind pedestrians walked on that surface, sometimes lining up to cross the roundabout at the assumed “corner” leading across the circulatory roadway, or alongside the circulatory roadway, but not at the crosswalk. Sidewalks at back of curb required blind pedestrians to follow the curb line with their cane, walking dangerously (and uncomfortably) close to traffic along the curving sidewalk, searching for the slope of a curb ramp, detectable warnings or other indication of the crosswalk. On islands, if the area outside the cut-through walkway was paved, some participants stepped up onto the island and were confused when they contacted the cut-through area, thinking the cut-through was the street rather than a pedestrian walkway.

NCHRP 3-78b: Final Project Report April 2016 56 Figure 4-23: Pavers, with fairly rough texture (cobblestone) This figure shows a photo of a roundabout approach with surface material that was not recognized as a non‐walking surface by blind participants. The cobblestone surface was installed at this roundabout between the concrete paved sidewalk and the curb, but it did not provide guidance (that might have been intended). Inset on right shows size of cobblestones in comparison to a person’s foot; each cobblestone is approximately the width of the foot, with an inch or more of grout between stones. Figure 4-24: Sidewalk at back of curb with no landscape separation Photo of a 5-foot-wide sidewalk that is right behind the curb, with a guard rail on the back side of the sidewalk. Crosswalk is visible ahead and a bus is in the lane close to the sidewalk, leaning toward the sidewalk. Blind pedestrians have to follow the curb line with their cane to find the detectable warning surface and curb ramp in order to locate the crosswalk. Does not provide separation required by proposed PROWAG

NCHRP 3-78b: Final Project Report April 2016 57 Figure 4-25: Paved colored surfaces do not provide adequate cues Photo shows a portion of an island where reddish color was added to the island pavement outside the white concrete pedestrian pathway between the crosswalks. Blind pedestrians did not walk within the concrete pathway through island and did not consistently locate the proper clossing location for the crossing from the island. At locations with parallel curb ramps, it appeared that participants were unfamiliar with this type of ramp design. Parallel curb ramps are most often used where there is limited right-of-way and the sidewalk is at back of curb. The entire sidewalk slopes down to a level landing, or turning space, where the crosswalk is located. Some blind participants stopped on the slope, parallel to the crossing, thinking that they had reached an edge of the street when they reached the level landing. This slope and the need to turn 90 degrees to cross were confusing to them. Some blind participants mistook the curb at the back of the ramp for the edge of the street and attempted to cross from behind or on top of the curb. Alignment cues are limited and the landing is often level with the street. Parallel curb ramps are described and shown in Figure 4-26.

NCHRP 3-78b: Final Project Report April 2016 58 Figure 4-26: Parallel ramp, showing curb at back of ramp Photo shows a parallel curb ramp at a roundabout crossing. There is no landscaping or barrier between the sidewalk and the curb. The entire 5-foot wide sidewalk slopes down to a level landing at the crosswalk location. The detectable warning surface is installed along the curb line for the width of the level area. There is a curb at the back of the sidewalk behind the landing to keep dirt and gravel from washing onto the landing area. 4.2.3 Detecting the Street Another aspect of locating the crosswalk is recognizing the street edge. At most locations, detectable warning surfaces were installed to indicate the edge of the street at curb ramps. The detectable warning surface must extend the full width of area that is level with the street to provide an adequate warning. At the roundabout and CTL locations where detectable warnings were not installed or did not extend the full width of the level area, some research participants continued into the street without recognizing that they had walked into the street. This occurred at locations where there was no detectable warning surface and where the detectable warning did not extend the full width of the cut-through area on the islands. Note that there is a considerable body of research regarding the detectability of underfoot surfaces by persons who are blind, either using their foot or the long cane for detection. Surfaces used in some locations, such as the scored concrete shown in Figure 4-27, are not among the surfaces that have been shown in research to be reliably detectable.

NCHRP 3-78b: Final Project Report April 2016 59 Figure 4-27: CTL with raised crosswalk and no detectable warning surfaces. Scored concrete is not detectable. Photo shows a raised crosswalk leading to an island. There are no detectable warning surfaces to indicate the edge of the street, either on the curb or the island. The concrete sidewalk is scored with approximately 12-inch squares, but as noted in text, that scoring is not detectable under foot or with a cane. Crosswalk markings in this photo are also very faded. At locations with raised crosswalks where changes in elevation between street and curb ramp are not present, the detectable warning surface is the only clue to the location of the edge of the street. If detectable warnings were not installed, participants were likely to walk into the vehicular lanes (Figure 4-27). However, even where detectable warnings were installed, participants walked into the street if they approached from an angle that allowed them to continue without stepping off the curb or actually crossing the detectable warning surface (Figure 4-28).

NCHRP 3-78b: Final Project Report April 2016 60 Figure 4-28: View of raised crosswalk where some participants did not find detectable warnings Photo of a raised crosswalk at a CTL where detectable warnings were installed, but there was a level area on each side of the crosswalk where a pedestrian approaching at an angle could walk in to the street without contacting the detectable warning surface. A participant walking in the direction shown by the arrow could walk into the street without contacting the detectable warning surface because the sidewalk was level with the street where the arrow is pointing Detecting the street is also important at both edges of both triangular islands at CTLs and splitter islands at roundabouts. Islands with cut-through walkways and no detectable warnings at each edge of the island were undetected by blind participants at both roundabouts and CTLs during the wayfinding studies. Where the island was not detected, participants usually failed to stop before entering other lanes of the roadway. If the island is intended as a refuge, and if pedestrians are expected to stop and consider the traffic on the other portion of the roadway before continuing, detectable warning surfaces must be installed for the entire width of the cut-through area and should extend at least 24” in the direction of travel at the edge of the island on each side of the island. At locations where the islands are not close to the crosswalk edges, as shown in Figure 4-29, blind pedestrians did not recognize that there was a potential refuge area, while sighted pedestrians made two-stage crossings, pausing at the islands to wait for gaps in traffic.

NCHRP 3-78b: Final Project Report April 2016 61 Figure 4-29: Roundabout without detectable surfaces or refuge in splitter islands. Photo of a crosswalk where the raised island portions do not extend to the edges of the crosswalk and there is also no detectable warning surface at the island location to delineate the edges of the island to blind pedestrians. Islands were not detectable. This resulted in pedestrians needing to make a full crossing rather than being able to make a 2 stage crossing Comments from participants also indicated that they felt more confident detecting the street where there was a slope difference as well as the detectable warning surface. They generally preferred islands that had a curb ramp and raised island rather than where the walkway was cut through at the level of the roadway. For wayfinding through the island with curb ramps, participants performed best with a paved walkway area and grass or gravel surfaces outside the walkway area. 4.2.4 Aligning to Cross It is important that individuals align themselves when preparing to cross a street so that they are facing more or less toward the curb ramp on the opposite side of the street. If they are significantly misaligned, they may walk toward or away from the intersection, and sometimes may become disoriented. Walking more or less directly across may be challenging for travelers who rely only on nonvisual cues. When aligning to cross, blind pedestrians used a combination of cues, including underfoot surfaces and traffic movement. The direction of vehicular traffic across the crosswalk, the alignment of the approach sidewalk, the detectable warning surface, the slope of the ramp, and the gutter/edge of street are all cues that may be used by individuals who are blind. While none of these cues are consistently oriented parallel or perpendicular to the direction of travel on the crosswalk (regardless of the type of intersection to be negotiated), and none of them are sufficient to enable accurate alignment (even if they are aligned with the crosswalk), they all contribute to alignment decisions. At locations where all cues named above provided the same information, individuals were more likely

NCHRP 3-78b: Final Project Report April 2016 62 to begin their crossing aligned within the crosswalk. Where there were mixed messages regarding the direction of travel, various techniques and cues were used across participants and across times of day. If traffic was heavy, the direction of vehicular traffic across the crosswalk seemed to influence decisions more than at times when traffic was light. Despite some limitations of the wayfinding component of this project in terms of site selection and sample size (discussed elsewhere in this report), the descriptive data of Appendix A suggest that aligning to cross is a major problem at most roundabouts and CTL’s. Unlike the previously discussed data for determining the crossing location, there were very few sites or conditions where participants consistently aligned successfully, and the alignment failure rates tended to be very large. While a few design features appeared to help in aligning to cross, as discussed below, the overwhelming weight of the observational and descriptive evidence suggests that blind pedestrians align poorly at roundabouts and CTL’s. For example, omnibus Figure 14-1 in Appendix A shows that, overall, initial alignment error rates averaged 34% at roundabouts and 43% at CTL’s. For the micro level of individual sites, Table 14-4 shows very high rates of initial alignment error at all but one of the CTL’s. This pattern of consistent and high levels of error is evident across all of the relevant Appendix A tables and figures. While our results invite stronger scrutiny of the problem with more focused experiments, they unfortunately do not provide much evidence that the particular crosswalk designs evaluated in this project promote alignment success. It may be the case that additional design features, perhaps similar to the prototype features tested in other work by our team (described elsewhere in this report), will ultimately prove necessary to enable successful initial alignment. 4.2.4.1 Features that appeared to aid in aligning to cross More accurate alignment decisions appeared to be made at locations where the approach to the crosswalk was aligned with the crosswalk, and where there were also returned curbs aligned with the direction of travel on the crosswalk. In addition, our observations the blind participants’ stepping movements in seeking an alignment direction suggested they sometimes used the gutter of the street, detectable warning surface, and the slope of the ramp for alignment. Although individuals stated that they did not line up with the ramps, and research indicates that people who are blind do not accurately line up perpendicularly to ramp slope, many did appear to the observing O&M specialists to be adjusting their alignment direction relative to the combination of the ramp slope and the gutter orientation. In most cases, the placement of the crosswalk can make a difference in the alignment of these features. For example, the base of curb ramps must be perpendicular to the gutter to avoid tipping by wheelchair users, so the location and angle of the ramp can be affected by the location on the curve of the roundabout or CTL roadway. Aligning all these features may require moving the crosswalk to a different point.

NCHRP 3-78b: Final Project Report April 2016 63 Figure 4-30: Approach direction, landscaping, returned curbs, detectable warning surface and crosswalk are all aligned, providing potential alignment cues for blind pedestrians Photo shows a roundabout crosswalk where the approach, the edges of the landscaping, the detectable warning surfaces, and the gutter are generally aligned with the crosswalk direction. 4.2.4.2 Features that seemed to provide inadequate or confusing information to pedestrians who were blind in aligning to cross When the various features mentioned above (the landscaping or edge of the sidewalk, the curb ramp slope, detectable warning surface, and edge of the street/gutter) were not aligned with the crosswalk direction, participants appeared more likely to be misaligned. In the examples shown in Figure 4-31 and Figure 4-32, the crosswalk angled slightly to the right, while the other cues were aligned toward the left, which appeared to cause confusion. This geometric situation is seen fairly commonly when a crosswalk is close to the roundabout’s circulatory roadway, since the curb ramp must intersect the gutter at right angles to be accessible to wheelchair users and avoid a tipping hazard. This was also noted at several CTL crosswalks.

NCHRP 3-78b: Final Project Report April 2016 64 Figure 4-31: Landscaping, curb ramp slope, detectable warning surface and gutter are not aligned with direction of travel on the crosswalk. This appeared to lead to alignment errors by blind pedestrians Photo of a curb ramp, crosswalk and island. The ramp, detectable warning surface and gutter all are aligned to the left of the crosswalk direction. The splitter island, which has a wider cut-through area than the crosswalk width, has detectable warning surfaces covering only a portion of the opening in the island. Figure 4-32: Ramp slope, detectable warning surface and gutter are all aligned to left of crosswalk direction Photo of a crosswalk from a CTL island across the main signalized lanes of traffic. The curb ramp, detectable warning and gutter are all aligned to the left of the crosswalk direction. Blind participants appeared to align with these features into the traffic lanes, as shown by arrow. Traffic cues were also confusing at this slightly skewed intersection

NCHRP 3-78b: Final Project Report April 2016 65 At one CTL location, where participants were observed to be aligning themselves with the slope of the ramp and the line of the gutter, participants routinely missed the small island as they crossed. Instead, they inadvertently entered the main intersection outside of any marked crosswalks. This situation is shown in Figure 4-33. Figure 4-33: At this location, some blind participants appeared to align themselves with the slope of the curb ramp and the line of the gutter, which led them to completely miss contacting the island. Photo of a wide sidewalk approaching a crosswalk and a relatively small island at a CTL. The crosswalk is generally aligned with the sidewalk approach direction, however slope of the ramp and the gutter or grade break is at an angle left of the crosswalk direction. Another feature of several locations was a cut-through area on the island that was narrower than the crosswalk, curb ramp, and landscaped area at the approach end of crosswalk. This situation can be problematic for large groups of sighted pedestrians, who must channelize themselves upon reaching the island. The issue for blind participants was largely one of confusion about their location as they contacted the island, but this was sometimes mitigated by island characteristics. For example, at the location shown in Figure 4-34 where there was a narrow cut-through area, with grass on other parts of the island, participants searched for the cut-through area and found it. At similar locations where there was a narrow cut-through area and a paved island, blind participants commonly stepped up onto the island and were confused and disoriented when they contacted the cut-through area, thinking that it was the street.

NCHRP 3-78b: Final Project Report April 2016 66 Figure 4-34: Cut-through is narrower than the crosswalk Photo of crosswalk at CTL. Crosswalk is approximately 10-feet-wide and cut-through area of the grassy island at the end of the crosswalk is approximately 5 feet. 4.2.5 Maintaining Correct Heading While Crossing and Staying Within the Crosswalk At traditional, rectilinear intersections, blind pedestrians who are initially misaligned may, as a result of the misalignment, begin their crossing in the wrong direction. And even those blind pedestrians who are initially well aligned may veer from their intended heading as they walk and may therefore leave the crosswalk while crossing. Importantly, at traditional intersections, this initial heading error is typically detected and corrected on the basis of acoustic information about the trajectory of traffic moving parallel to the crosswalk. Experienced blind pedestrians can hear whether they are walking toward, away from, or parallel to the traffic moving next to them. If they hear that they are walking toward or away from this traffic, they then adjust their trajectory to be parallel to the traffic trajectory and thereby complete their crossing at the appropriate ending location. However, at most roundabouts and CTL crossing locations, this “parallel traffic” is not present. Without such traffic, blind pedestrians have few, if any, environmental cues to help maintain a correct initial heading or to make corrections to an incorrect initial heading. Short crossings allow less opportunity to veer from the initial heading; however, so shorter crossing distances (i.e. narrow lanes) may be an advantage. The research team expected raised crosswalks to assist blind participants to stay within the crosswalk, based on the assumption that any heading error would be corrected when participants detected cross slope on either side of the crosswalk when they contacted it with their cane or feet. However, the team did not observe raised crosswalks to be as helpful as expected. We speculate that many of the blind participants did not understand the design of the raised crosswalk well enough to recognize the cross slope as a cue that they were veering out of the crosswalk. With wide raised crosswalks and short crossing distances, there also were very few instances of veering outside the crosswalk.

NCHRP 3-78b: Final Project Report April 2016 67 It is important to note that well-marked crosswalks can provide important information to assist pedestrians with low vision stay within the crosswalk. However, participants in this research were all individuals without usable vision, so no data was gathered on the effect of crosswalk markings. The Appendix A data for maintaining correct heading and staying within the crosswalk (which results in reaching the correct crossing ending location) were consistent with our observations, with all but the two roundabout sites in Ann Arbor, Michigan and one site in Cary, North Carolina showing high rates of failure to arrive at the crossing ending location. For example, Figure 14-1 shows that, overall, crossing ending location error averaged 18% at roundabouts and 38% at CTL’s. At the level of individual sites, Table 14-4 and Table 14-5 show that with the exception of the three roundabout sites just mentioned, there were typically high rates of crossing ending location error at roundabouts. For the CTL sites, the lowest percentage failure for crossing ending location was 17% (at two Tucson, Arizona sites) and the highest was 70% (at one Boulder, Colorado site and one Tucson site). While the need for rigorous follow-up to the descriptive wayfinding components of this project has been emphasized throughout this report, the widespread and high rates of our blind participants’ challenges in arriving at the appropriate ending location are unacceptable but somewhat expected. These are complex travel situations, and our participants typically had little or no experience with CTLs and roundabouts. However, widely accepted definitions of “accessibility” include access to intersections and other aspects of the public right-of-way that individuals are not familiar with, either at a general or intersection specific level. The challenges and associated task error rates we observed are unacceptable because no reasonable definition of “accessible” would include the task failure rates observed in our descriptive data, as preliminary as it is. But these data are also somewhat expected given that, with the exception of the raised crosswalks, there were few design features at the roundabouts and CTL’s that might be expected to enhance the detection and correction of heading error. As for the previously discussed problem of alignment error, it may be the case that additional design features, perhaps similar to prototype guidestrip features tested in other work by our team (described in Section 2.3 of this report), will ultimately prove necessary. The three roundabout sites where participants always succeeded in reaching the crossing ending location were the Nixon & Huron site in Ann Arbor (S entry and S exit – see Photo Log, Figures 12-155 through 12-157, and Figures 12-159 & 12-160), the Ellsworth & State site in Ann Arbor (W exit, Photo Log, Figures 12-149 through 12-151), and the Old Apex and Chantam site in Cary (W entry, Photo Log Figures 12-209 & 12-210). These crossings have several commonalities. First, the crosswalks are very wide (relative to the direction of pedestrian travel, “long” in the direction of vehicle travel), with most being nearly as wide as they are long. Second, these sites’ potential alignment cues (curb/gutter, curb ramp, and returned curbs) were aligned or only slightly misaligned with the crosswalk. Third, three of the four crossings are single-lane, resulting in a short crossing (relative to the direction of pedestrian travel). We hypothesize that at these locations, the participants were initially reasonably well aligned, and that the wide crosswalks combined with the short crossing distances meant that a substantial amount of misalignment or veering error could be tolerated before the participants would have left the crosswalk before completing crossings. Appendix A includes specific data about these sites. 4.2.6 Other Features – Pushbuttons, Particularly on Islands Pushbuttons for pedestrian signals, accessible pedestrian signals, RRFBs, or other devices must be located close to the crosswalk they control in order to be useful to pedestrians who are blind. Despite a locator tone on the accessible pedestrian signal shown in Figure 4-35, participants in the wayfinding study did not find the pushbutton for the crossing because it was located too far back from the street. If they searched for it, they turned around before they found it. Blind participants also had a problem with using the correct pushbutton for their crossing. Most of the pushbuttons at locations where data was collected were not audible/accessible. Participants commonly pushed the wrong pushbutton on the island.

NCHRP 3-78b: Final Project Report April 2016 68 Figure 4-35: The APS pedestrian pushbutton on this island was not found by blind participants. Photo taken from the center of a CTL island looking across a wide street with a marked crosswalk. In the foreground is a pole with an APS pushbutton on it, located in the grass about 2 feet to the right of the paved cut-through walkway leading to the crosswalk. The pole and pushbutton are approximately 15 feet from the edge of the street pictured, and the pushbutton locator tone was not audible more than 5 feet from the pushbutton. 4.2.7 Summary The primary features that appeared to enhance wayfinding at both roundabouts and CTLs included: • Grass or other landscape strip between sidewalk and curb; • Grass or gravel outside the crosswalk area, particularly on islands; • Having all wayfinding cues for alignment (vehicular traffic across the crosswalk, approach direction, landscaping or edge of the sidewalk, the curb ramp slope, detectable warning surface, and edge of the street/gutter) provide the same information; • Short crossings; • Detectable warnings to indicate the edge of the street and islands; and • Audible information devices or audible pedestrian signals that were located close to the crosswalk they control and properly oriented. These features may need to be considered in designing roundabouts and CTLs. Location of crosswalks and landscaping near the crosswalk and on the island can be key features that influence accessibility of

NCHRP 3-78b: Final Project Report April 2016 69 these types of intersections. Training of blind pedestrians cannot improve wayfinding in an environment without usable cues. Among the limitations of this early wayfinding study are differences in where participants began their trials (varied from 10 feet to over 100 feet from the crosswalk), differences in trial sample sizes across study locations, and differences in the wayfinding abilities of the participants (especially differences in ability between subjects who are typically dependent on dog guides for navigation and those who are not) that may act as confounds. Despite the recognized limitations of the wayfinding component of this project, the team believes that, taken together, the observational and descriptive findings suggest that wayfinding challenges pose serious problems to the accessibility of roundabouts and CTL’s. We urge policy makers and designers to support further research to better document, understand, and mitigate these challenges.

NCHRP 3-78b: Final Project Report April 2016 70 --- This page intentionally left blank ---