Cover Image

Not for Sale

View/Hide Left Panel
Click for next page ( 111

The National Academies of Sciences, Engineering, and Medicine
500 Fifth St. N.W. | Washington, D.C. 20001

Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement

Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 110
110 PART 2 Detailed Single-Lane Roundabout: Golden, CO Introduction ing itself is at pavement elevation. No pedestrian-detectable warning surfaces were installed on the splitter island and so This section describes detailed analysis results of data col- the study participants were instructed by the O&M specialist lected at the single-lane roundabout in Golden, CO, at the when they completed the first half of the crossing. Detectable intersection of Golden Road and Ulysses Road (Exhibit 13). warnings were installed on the outside curb ramps, and the The initial focus is on pedestrian-related measures, including crosswalk was outfitted with standard pedestrian signage. the availability and utilization of yield and gaps, as well as pedestrian delay and O&M interventions. The single-lane roundabout was studied twice, coinciding Crossing Statistics for Crosswalk with the pre and post studies at the nearby two-lane round- The analysis of crossing performance focuses on aspects about. But contrary to that site, no pedestrian crossing treat- of pedestrianvehicle interaction following the NCHRP ments were installed at this single-lane roundabout. Without Project 3-78A analysis framework. The first analysis compo- treatment installation, the prepost comparison serves as a nent describes the availability and utilization of yields in both control for any learning effects or changes in driver behavior the pre and post treatment conditions. Two yield measures between the two studies. The analysis presents findings in the are used in the analysis: pre and post conditions for the studied crosswalk sequentially. Only the eastern crosswalk was included in the study. The P(Y_ENC): The probability of encountering a yield event, results are also compared to those gathered at other single- defined as the number of yields divided by the total of all lane roundabouts included in this study. events encountered by the pedestrian until he/she completes The pre study was completed in July 2008 and had a total the crossing. of 18 blind travelers. Thirteen of the original 18 participants P(GO|Y): The probability of yield utilization, defined by returned for the post experiment in September 2008. Again, the number of crossings in a yield divided by total number no treatments were installed at this roundabout, so the under- of yields encountered by the pedestrian. lying hypothesis is that overall performance in pre and post conditions are the same. The P(Y_ENC) measure is somewhat different from the traditionally used probability of yielding, since it is calculated Golden Single-Lane Analysis Results on the basis of all pedestrianvehicle events and not just potential yielders. Chapter 3 provides additional discussion Site Description on these and other performance measures, including examples A picture of the studied crosswalk is shown in Exhibit 14. on the difference between the yielding measures. The roundabout has a central island diameter of 100 ft, includ- Exhibit 15 shows the statistics for the studied crosswalk. ing a 10-ft truck apron. The lanes at the studied crosswalk are The figures shown represent the mean results by crossing leg 20 ft wide, partly to accommodate a nearby roadside bus stop. considering all subjects. Each subject completed four crossing The crosswalk is located approximately 60 ft from the circu- trials at the roundabout, with each trial consisting of four lane lating lane measured at the exit side, and approximately 50 ft crossings (e.g., entryexitexitentry). For example, a subject from the roundabout yield line at the entry. The two-stage in the pre condition would have crossed the entry and exit crossing is divided by an 8-ft raised splitter island, but the cross- portions of the crosswalk, respectively, eight times (twice in

OCR for page 110
111 Exhibit 13. Aerial view of roundabout. Exhibit 15. Yield availability and utilization statistics for studied crosswalk. a) P(Y_ENC) Pre (n = 18) Avg. Min. Max. Std. Dev. Entry 51.1% 16.7% 100.0% 18.4% Exit 29.6% 7.1% 57.1% 13.7% Overall 40.4% 7.1% 100.0% 19.4% Post (n = 13) Entry 51.1% 18.8% 100.0% 21.1% Exit 36.5% 13.6% 62.5% 16.5% Overall 43.8% 13.6% 100.0% 20.0% b) P(GO|Y) Pre (n = 18) Avg. Min. Max. Std. Dev. Entry 82.8% 36.4% 100.0% 20.1% Photo by Google Exit 76.0% 25.0% 100.0% 26.1% Overall 79.4% 25.0% 100.0% 23.2% Post (n = 13) Entry 80.3% 21.3% 100.0% 23.3% Exit 89.2% 66.7% 100.0% 12.7% Overall 84.7% 23.1% 100.0% 18.9% each of four trials). The average performance for the entry leg in the pre condition is then calculated from the mean of these significantly in the pre and post studies, suggesting that driver eight crossings for all 18 subjects. The overall average is then behavior was comparable between the two studies. calculated from 36 observations (18 entry and 18 exit) each The exhibit further shows the rates of yield utilization, representing eight individual crossing attempts. In the post P(GO|Y), defined as the rate of yields that resulted in a pedes- conditions, 13 subjects participated in the experiment. In total, trian crossing the roadway. The yield utilization rates are gen- 31 subjects were included in the study (18 pre, 13 post) and erally on the order of 75% to 90%, and no significant differences each performed 16 lane crossings (four trials at four lanes each), were observed in either the prepost or entryexit comparisons resulting in a total of 496 crossing attempts at this location. due to large standard deviations in the mean estimates. Exhibit 15 shows that the probability of encountering a A considerable fraction of yields further fell into the "forced yield, P(Y_ENC), was higher in the entry lane than in the yield" category, which is defined as the pedestrian stepping exit lane for both the pre (p = 0.0004) and post (p = 0.0624) out into the roadway before the vehicle initiated the yielding conditions. The yield encounter probability did not change process. The degree of risk associated with these events depends on the relative position and speed of the vehicle at the time of crossing initiation. Forced yield events should therefore not Exhibit 14. The studied crosswalk. necessarily be interpreted as poor or risky decisions. In the pre condition, 32.5% and 40.6% of yields were forced at the entry and exit leg, respectively. In the post condition, the cor- responding forced yield percentages were 22.2% and 39.4%. The differences between pre and post percentages of forced yields are not statistically significant (p = 0.2294 and p = 0.8955 for entry and exit, respectively). The exit leg crossing did show a greater percentage of forced yields for both pre and post conditions, but these differences were also not statistically significant. Photo by Janet Barlow The analysis next considered the availability and utilization of crossable gaps. For the purpose of this analysis, a crossable gap was defined as a gap greater than 8 s which was sufficient to cross the wide 21-ft crosswalk at a walking speed of 3.5 ft/s, while allowing for a 2-s safety buffer. This 2 s allows for some pedestrian reaction time before initiating the crossing, as well

OCR for page 110
112 as a safety buffer between a completed crossing and the next The combined effect of gap and yield availability and utiliza- vehicle arrival. Similar to the yield statistics, two gap-related tion is reflected in the delay experienced by pedestrians. Delay parameters are defined: statistics in Exhibit 17 are provided for two delay measures: P(CG_ENC): The probability of encountering a CG event Observed Delay per Leg (s): The average pedestrian delay (gap greater than 8 s), defined as the number of crossable in seconds, defined as the time difference between when the gaps divided by the total of all events encountered by the trial started and when the pedestrian initiated the crossing. pedestrian. Delay>Min (s): The delay beyond the first opportunity P(GO|CG): The probability of crossable gap utilization, (Delay>Min), defined as the time difference between first defined by the number of crossings in a CG divided by total yield or crossable gap encountered by the pedestrian and number of CGs encountered by the pedestrian. the actual crossing initiation. Exhibit 16 shows the statistics for the studied crosswalk. Statistics for all measures are for crossing one leg of the The results in Exhibit 16 show that the P(CG_ENC) is slightly roundabout at either the exit or entry approach. The total higher on the entry leg for the pre conditions, but this difference average delay by crossing can be calculated by summing delay is not significant at p = 0.1125. There is no significant effect statistics for the entry and exit legs. of P(CG_ENC) in a prepost comparison, suggesting that Exhibit 17 shows that small differences in average pedestrian traffic patterns with respect to gap availability remained largely delay per leg were observed in an entryexit leg comparison, unchanged between the two studies. Exhibit 16 further shows that the blind study participants although none of the small differences were statistically signifi- generally had high crossable gap utilization rates, averaging cant at the given sample size (all p > 0.40). Also, no significant in the 80% to 90% range. This may be the result of a very con- differences in delay were observed in a prepost comparison servative crossable gap definition that allows most pedestrians (all p > 0.50). In addition to the average delay for all participants, to cross. In fact, some assertive pedestrians crossed in gaps it is important to emphasize that some participants experienced smaller than this threshold, resulting in P(GO|Gap>Min) much larger delays. The longest overall average delay by a values greater than 100% (capped in Exhibit 16). No significant participant was 51.4 s per leg. A 2-min time-out was used for difference in P(GO|CG) is detected in a prepost or entryexit all trials, but none of the participants ever reached that limit comparison. at this site. The results for Delay>Min also show no significant difference between for prepost and entryexit comparisons. Overall, Exhibit 16. Crossable gap availability and utilization the Delay>Min results suggest that the blind pedestrians did not statistics for studied crosswalk. a) P(CG_ENC) Exhibit 17. Average pedestrian delay statistics Pre (n = 18) Avg. Min. Max. Std. Dev. for studied crosswalk. Entry 26.3% 0.0% 44.4% 12.4% Exit 20.6% 4.8% 34.8% 8.4% a) Observed Delay per Leg (s) Overall 23.5% 0.0% 44.4% 10.8% Pre (n = 18) Avg. Min. Max. Std. Dev. Post (n = 13) Entry 10.9 4.0 31.3 7.3 Entry 21.5% 0.0% 37.5% 9.4% Exit 13.0 3.5 29.4 7.9 Exit 21.1% 8.3% 50.0% 11.3% Overall 11.9 3.5 31.3 7.6 Overall 21.3% 0.0% 50.0% 10.2% Post (n = 13) b) P(GO|CG) Entry 13.3 2.7 51.4 13.6 Pre (n = 18) Avg. Min. Max. Std. Dev. Exit 11.0 3.4 27.8 7.3 Entry 83.2% 33.3% 100.0%* 23.7% Overall 12.1 2.7 51.4 10.7 Exit 86.8% 40.0% 100.0%* 23.4% b) Delay>Min (s) Overall 85.1% 33.3% 100.0%* 23.2% Pre (n = 18) Avg. Min. Max. Std. Dev. Post (n = 13) Entry 2.8 0.1 6.5 2.1 Entry 80.9% 45.5% 100.0% 22.3% Exit 2.7 0.1 7.0 2.3 Exit 81.4% 40.0% 100.0% 22.5% Overall 2.8 0.1 7.0 2.2 Overall 81.2% 40.0% 100.0% 21.9% Post (n = 13) * These figures were capped at 100%, although the calculation resulted in Entry 3.7 0.3 19.9 5.2 estimates greater than 100%. This occurs if pedestrians utilize some Exit 2.5 0.1 9.9 2.8 non-crossable gaps and therefore have more utilized gaps than there are Overall 3.1 0.1 19.9 4.2 crossable gaps available.

OCR for page 110
113 Exhibit 18. Delay distribution all subjects single-lane roundabout. Single-Lane Roundabout 100 PRE 90 22.2 sec. 85%ILE POST DELAY 80 21.7 sec. 70 60 Percentile 50 40 30 PRE 20 POST 10 0 0 10 20 30 40 50 60 70 80 90 100 110 120 Delay (sec.) miss a lot of crossing opportunities. Despite these low averages, Exhibit 19 shows the 85th percentile delay estimate by subject. some pedestrians experienced Delay>Min up to 19.9 s. The null It appears that one participant (subject 7) experienced sig- hypothesis that Delay>Min = 0 is rejected for both pre and nificantly greater delay in the post condition, while all other post conditions (p < 0.0001 and p = 0.0030, respectively). delay performances remained largely unchanged. The delay Exhibit 18 shows the distribution of delay for all subjects in statistics are arranged by time of day during which subjects par- the pre and post conditions. The hypothesis that no significant ticipated, but no trends can be identified. Note that subjects 1, changes took place between the studies is supported by the 5, 10, 15, and 16 did not participate in the post study. data. This implies that any effects observed at the neighboring The analysis further investigated two new parameters that two-lane roundabout are likely attributable to the installation were not previously used in Schroeder, Rouphail, and Hughes of the treatments and not a learning effect by pedestrians. (2009). Both measures are intended to describe the efficiency Exhibit 19. 85th percentile delay by subject single-lane roundabout. Single-Lane Roundabout 120 PRE 85th Percentile Delay (sec.) 100 POST 80 60 40 20 0 1 3 6 9 11 16 4 7 10 12 14 17 2 5 8 13 15 18 9:00am 11:30am 3:30pm Subject

OCR for page 110
114 with which a crossing opportunity is utilized for both gaps ipants if the crossing decision would have resulted in undue and yields: risk to pedestrian and/or driver. The resulting rate of O&M intervention is defined as follows: Latency (s): Latency is defined as the time between when the last vehicle went through the crosswalk and the time Intervention Rate (%): The intervention rate is defined by the pedestrian initiated the crossing. the number of times the O&M specialist intervened for a Yield Lost Time (s): The YLT is defined as the time between particular subject divided by the total number of lanes when a driver first yields and the time the crossing is ini- crossed for a particular condition. For example, one inter- tiated. Note that in some cases, pedestrians may prefer to vention over a set of eight lane crossings at the roundabout cross only after a car has come to a full stop (stopped yield), entry corresponds to an intervention rate of 12.5%. and so some inherent yield utilization time is expected. The summary statistics for O&M Interventions are given Exhibit 20 shows statistics for both measures. in Exhibit 21. The latency results in Exhibit 20 suggest that on average The results show that a total of four O&M interventions pedestrians wait 4 to 6 s into a crossable gap before initiating were observed in the pre case over a total of 72 lane crossings the crossing, suggesting inefficiency in decision-making. No at the entry leg for a rate of 2.8%. In the post analysis the inter- significant difference in latency was detected for a prepost or vention rate at the entry was 1.0%; however, due to high stan- entryexit comparison. dard deviations that difference is not significant (p = 0.2616) For the yield lost time measure, pedestrians on average at this sample size. No interventions were observed at the exit wait 1.5 to 2 s before crossing in front of a yielding vehicle. leg of this particular roundabout. The overall intervention rates The average maximum YLT was 11.1 s for the pre and 6.6 s for for pre and post were 1.4% and 0.5%, respectively. Again, this the post condition. Note that in many cases, drivers will not be difference is not statistically significant. The null hypothesis that willing to wait this long and a high YLT will therefore translate the intervention rate is zero is rejected for the pre condition to an increased percentage of missed yields [lower P(GO|Y)]. (p = 0.0344) but cannot be rejected at the given sample size Also note that some YLT values are negative, suggesting that for the post evaluation. some pedestrians forced vehicles to yield. Exhibit 22 explores the distribution of interventions by sub- Finally, the analysis includes the rate of O&M interventions ject and time of day. Given the rare occurrence of interventions, that represent a measure of pedestrian risk during the crossings. it is difficult to draw any conclusions about patterns at the The study participants were at all times accompanied by a given sample size. Participants who didn't return for the post certified O&M specialist who was directed to stop the partic- study are shown with negative intervention rates to visually distinguish them from zero-intervention subjects. The num- Exhibit 20. Latency and yield lost time statistics bers are shown as the percentage of interventions from 16 lane for studied crosswalk. crossing per subject at this site. While the rates of interventions appear low, this does not a) Latency (s) mean that a crossing is safe. Ashmead et al. (2005) posited Pre (n = 18) Avg. Min. Max. Std. Dev. that the probability of a dangerous crossing decision is given Entry 5.7 1.8 16.0 4.0 by 1(1pper crossing)n, where pper crossing is the observed intervention Exit 4.9 2.4 10.5 1.9 rate and n the number of crossing attempts. After 40 crossings Overall 5.3 1.8 16.0 3.0 (twice per day, 5 days a week, over 4 weeks), the probabilities Post (n = 13) Entry 5.0 1.3 12.6 3.1 Exit 4.7 2.7 7.3 1.6 Exhibit 21. O&M intervention statistics for Overall 4.8 1.3 12.6 2.4 single-lane roundabout crosswalk. b) Yield Lost Time (s) Intervention Rate Pre (n = 18) Avg. Min. Max. Std. Dev. Pre (n = 18) Avg. Min. Max. Std. Dev. Entry 2.6 2.1 11.1 3.8 Entry 2.8% 0.0% 12.5% 5.3% Exit 0.5 1.9 5.1 2.0 Exit 0.0% 0.0% 0.0% 0.0% Overall 1.6 2.1 11.1 3.2 Overall 1.4% 0.0% 6.3% 2.7% Post (n = 13) Post (n = 13) Entry 2.8 0.1 6.6 2.3 Entry 1.0% 0.0% 12.5% 3.5% Exit 1.1 1.4 4.4 1.8 Exit 0.0% 0.0% 0.0% 0.0% Overall 1.9 1.4 6.6 2.2 Overall 0.5% 0.0% 6.3% 1.7%