Cover Image

Not for Sale



View/Hide Left Panel
Click for next page ( 125


The National Academies | 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 124
124 Exhibit 35. Nearfar lane effects, post condition, for RCW. Far-Lane Event Near-Lane Lane Rolling Stopped Forced X-Able Non-X. Multiple Events Event Outcome Yield Yield Yield Gap Gap RY STY CG FY non-CG Total Rolling Yield Utilized 4 5 6 6 0 2 1 2 1 0 27 Non-Utlz. 2 0 0 0 3 0 0 0 0 0 5 Stopped Yield Utilized 12 28 12 18 1 0 4 2 4 0 81 Non-Utlz. 0 2 0 0 3 0 0 0 0 1 6 Forced Yield Utilized 1 5 0 9 0 0 1 1 2 0 19 Non-Utlz. 0 0 0 0 0 0 0 0 0 0 0 Crossable Gap Utilized 5 11 13 35 1 1 0 7 2 0 75 Non-Utlz. 0 0 0 0 3 0 0 0 0 2 5 Non-Cross. Gap Utilized 0 1 0 1 3 0 0 0 0 0 5 Non-Utlz. 0 1 0 4 37 0 0 0 0 4 46 Total 24 53 31 73 51 3 6 12 9 7 269 Exhibit 36. Availability and utilization statistics and 21.6% of events were correctly rejected events. Only 1.5% for post condition, RCW. were classified as missed opportunities and inefficient behav- ior, and 2.6% fell into the "unsafe" category. Also, 21.2% of Post (n = 269) Near Lane Far Lane events were associated with a forced yield and were labeled as Availability Statistics inconclusive. P(Y_Enc) 51.3% 46.8% A comparison of Exhibits 34 and 37 makes evident that the P(CG_Enc) 29.7% 31.6% biggest difference between the pre and post data is a drastic Utilization Statistics reduction of rejected opportunities (reduced from 443 to 62 P(GO|Y) 92.0% 96.0% events). With the introduction of the raised crosswalk, drivers P(GO|CG) 93.8% 95.3% tended to yield more frequently, and many of these yields resulted in crossings. Furthermore, pedestrians seemed more comfortable accepting gaps, resulting in an overall drop of a crossable gap from all events was 29.7%, which is similar to inefficient decisions from 6.0% to 1.5%. As a result of this the pre study (25.5%). However, the rate of crossable gap more assertive behavior, the rate of potentially risky events utilization increased from 78.9% to 93.8% in the near lane. about doubled, as did the rate of inconclusive events. Note Exhibit 36 shows the summary availability and utilization that any O&M interventions were removed from the dataset statistics. prior to analysis (discussed separately) and so none of these Exhibit 36 shows that the availability of yield crossing forced yields resulted in a truly dangerous situation. opportunities about doubled with the installation of the raised crosswalk, while the availability of gap crossing oppor- Performance Statistics for RCW tunities remained largely unaffected. However, the rate of uti- lization of both types of opportunities increased drastically, The installation of the RCW is expected to also affect the with utilization rates well above 90%. Overall, far fewer events bottom-line delay and risk performance statistics for the pedes- were (had to be) rejected by the pedestrian, as is evident in the trians. Delay statistics in Exhibit 38 are provided for pedestrian summary statistics in Exhibit 37. delay in seconds, defined as the time difference between when Exhibit 37 shows that for the total of 269 events, 53.2% of the trial started and when the pedestrian initiated the crossing. crossings were correct utilizations of crossing opportunities The exhibit further shows the delay beyond the first oppor- Exhibit 37. Summary of pedestrian behavior, post condition, RCW. Pedestrian Crosswalk Condition Decision Crossable/Safe Non-Cross./Unsafe Inconclusive GO 143 53.2% 7 2.6% 57 21.2% NoGO 4 1.5% 58 21.6%

OCR for page 124
125 Exhibit 38. Average pedestrian delay statistics from 17.0 s to 6.7 s (p = 0.0434). There was no significant dif- for RCW. ference between the delay experienced at the entry and exit a) Observed Delay per Leg (s) portions of the crossing in the pre study. In the post study, the Pre Avg. Min. Max. Std. Dev. delay difference of 2.7 s higher average delay at the exit is sig- Entry (n = 18) 15.6 1.5 57.1 15.9 nificant at p = 0.0440. In addition to the average delay for all Exit (n = 18) 18.4 3.0 84.9 19.4 participants, it is important to emphasize that some participants Overall (n = 36) 17.0 1.5 84.9 17.6 experienced much larger delays. The highest average delay was Post 84.9 s in the pre study, and the single highest delay experienced Entry (n = 13) 6.7 3.6 12.2 2.7 by a study participant was 115.8 s (not shown). These figures do Exit (n = 13) 9.4 4.0 18.2 3.7 not include trials that were beyond the 2-min time-out limit. Overall (n = 26) 8.0 3.6 18.2 3.5 The maximum average delay in the post was only 11.2 s, and the b) Delay>Min (s) single highest delay in the post condition was 57.4 s. Overall, Pre Avg. Min. Max. Std. Dev. the 2-min time-out limit was reached 9 times for all subjects Entry (n = 18) 3.1 0.0 11.1 3.6 in the pre condition and never after installation of the raised Exit (n = 18) 3.8 0.2 11.2 2.8 crosswalk. With installation of the RCW, the observed range Overall (n = 36) 3.4 0.0 11.2 3.2 and standard deviation of average delay were reduced, sug- Post gesting more consistent behavior across subjects. Entry (n = 13) 1.7 0.1 4.3 1.3 The results for Delay>Min also show a reduction between Exit (n = 13) 2.8 0.2 5.8 1.7 the pre and post conditions from 3.4 to 2.3 s, but this differ- Overall (n = 26) 2.3 0.1 5.8 1.5 ence was not statistically significant (p = 0.2117). Overall, the Delay>Min results suggest that the blind pedestrians did not miss a lot of crossing opportunities, but rather were delayed tunity (Delay>Min), which was defined as the time difference due their infrequent occurrence. Despite these low averages, between the first yield or crossable gap encountered by the some pedestrians experienced Delay>Min up to 48.4 s in the pre pedestrian and the actual crossing initiation. Statistics for all condition and up to 36.9 s the post case (not shown). The max- measures are for crossing one leg (two lanes) of the round- imum average Delay>Min were 11.2 and 5.8 s, respectively. about. The statistics shown are calculated from the average Exhibit 39 shows the cumulative distribution of pedestrian crossing performance for each subject. The total sample size delay at the PHB. The 85th percentile delay is highlighted. was 18 and 13 subjects in the pre and post studies, respectively. The exhibit clearly shows a shift of the delay distribution, Each data point represents the average of 16 approach cross- with pedestrians in the post condition overall experiencing ings, half at the entry and half at the exit of the roundabout. lower delays. The 85th percentile delay across all participants Exhibit 38 shows that the average pedestrian delay per leg was reduced from 31.0 to 13.4 s. The difference is also evi- decreased significantly between the pre and post conditions, dent when looking at the crossing performance of individual Exhibit 39. Cumulative delay distribution of pedestrian delay at RCW. RCW 100 POST 90 13.4 sec. 80 PRE 85%ILE 31.0 sec. DELAY 70 Percentile 60 50 40 30 PRE 20 POST 10 0 0 10 20 30 40 50 60 70 80 90 100 110 120 Delay (sec.)

OCR for page 124
126 Exhibit 40. 85th percentile delay by subject Exhibit 41. Latency and yield lost time statistics for RCW. north crosswalk. RCW a) Latency (s) Pre Avg. Min. Max. Std. Dev. 120 Entry (n = 18) 6.5 1.3 15.3 4.0 85th Percentile Delay (sec.) PRE 100 Exit (n = 18) 8.1 1.9 41.4 8.7 POST Overall (n = 36) 7.3 1.3 41.4 6.7 80 Post 60 Entry (n = 13) 4.5 2.3 6.8 1.5 Exit (n = 13) 6.1 2.2 11.2 2.8 40 Overall (n = 26) 5.4 2.2 11.2 2.4 20 b) Yield Lost Time (s) Pre Avg. Min. Max. Std. Dev. 0 Entry (n = 18) 3.5 0.8 9.0 2.5 1 3 6 9 11 16 4 7 10 12 14 17 2 5 8 13 15 18 Exit (n = 18) 0.8 4.3 8.9 3.3 Overall (n = 36) 2.2 4.3 9.0 3.2 9:00am 11:30am 3:30pm Post Subject Entry (n = 13) 3.4 1.2 6.2 1.3 Exit (n = 13) 2.4 0.5 3.6 0.9 Overall (n = 26) 2.9 0.5 6.2 1.2 participants. Exhibit 40 shows the 85th percentile delay for all participants in the pre and post condition. Note that partic- ipants 1, 5, 10, 15, and 16 did not participate in the post study. will therefore translate to an increased percentage of missed The figure shows that the 85th percentile delay was reduced yields [lower P(GO|Y)]. Note that the minimum YLT is neg- for every participant in the post condition but also that the ative, suggesting that some pedestrians forced vehicles to effect was greatest for those that experienced high delay in the yield (the yield was "utilized" before it occurred). The instal- pre. So, in addition to reducing the overall delay, the RCW lation of the RCW had no significant effect on yield lost time. also created a more uniform distribution of delay, even for The above measures primarily focus on the efficiency of participants with presumably worse travel skills. Exhibit 40 crossing, and largely ignore the risk experienced by pedestri- further explores the relationship between 85th percentile delay ans. While delay and other efficiency measures are used fre- and the time of day of the participation. From a visual analysis, quently by engineers, they fail to capture the human element no trends are observed. of crossing risk. The selected surrogate risk measure for this The analysis further investigated two parameters intended study is the number of times the O&M specialist had to inter- to describe the efficiency with which a crossing opportunity vene in the crossing. Exhibit 42 shows the frequency and rate is utilized. For utilized gaps, the latency is defined as the time of O&M interventions for all trials. between when the last vehicle went through the crosswalk and Exhibit 42 shows a drastic reduction in the occurrence of the time the pedestrian initiated the crossing. For utilized interventions. The percentage of trials that resulted in an yields, the yield lost time is defined as the time between when a O&M intervention is reduced from 2.8% to zero in the post driver first slows down for a yield and the time the crossing is condition. Following discussion in Ashmead et al. (2005), a initiated. Note that in some cases, pedestrians may prefer to 2.8% likelihood of a risky decision will result in a cumulative cross only after a car has come to a full stop (stopped yield) and risk of 67.9% after 40 crossings (for example two crossings a so some inherent yield utilization time is expected. Exhibit 41 day over 4 weeks with 5 working days per week). In the pre shows statistics for both measures. case, the exit lane had a slightly higher intervention rate than The latency results in Exhibit 41 suggest that on average the entry, which is consistent with findings at other multi-lane pedestrians wait 7.3 s into a crossable gap before initiating the roundabouts. However, given that interventions are very rare crossing, suggesting inefficiency in decision-making. With the events, it is unlikely that the post intervention is an absolute installation of the RCW, the average latency decreases slightly zero, but rather small-enough to where it was not measura- to 5.4 s (p = 0.2767). ble at the given sample size. Exhibit 43 explores the distribu- For the YLT measure, pedestrians in the pre condition lose tion of O&M intervention across subjects and by time of day. an average of 2.2 s before crossing in front of a yielding vehicle. Subjects that didn't participate in the post experiment are However, the average maximum YLT was 9.0 s. In many cases, shown with a negative intervention rate to distinguish them drivers will not be willing to wait this long, and a high YLT from those with zero interventions that did participate.