TRCLC 17-1

An intersection database enhances blind pedestrians鈥 access to complex signalized intersections: Stage 2 analysis & database development

PIs: Dae Kim Ph.D., David Guth, Ph.D, Paul Ponchillia, Ph.D., Janet Barlow, Lee Rodegerdts. 
Start and End Dates: 8/15/2017 鈥 8/31/2018
 

Summary:

This project investigated the effects of providing verbal descriptions of intersections and crosswalks on the performance of street-crossing subtasks by individuals who are totally blind. The intersections included crosswalks that varied widely in geometric and operational characteristics, including the presence or absence of accessibility features. In the no-database condition, participants used their typical street-crossing procedures. In the database-condition, participants additionally listened to database-generated descriptions of the intersections and crosswalks before crossing. The database descriptions had significant positive effects on some subtasks (primarily 鈥渃rossing鈥 subtasks such as deciding when to cross) and not others (primarily 鈥渨ayfinding鈥 subtasks such as remaining in the crosswalk).

 

Research Results:

Completing Crossing within Crosswalk Lines

The predictor variables were database information and APS. The interaction effect was not significant (p = 0.715). However, participants completed crossing within the crosswalk lines more frequently when the APS was present (mean = 54%, SE = 7.4%) than when it was not (mean = 14%, SE = 3.7%, odds ratio = 8.83, p = 0.003). That is, the odds of participants completing crossing within the crosswalk lines were almost nine times higher when the APS was present than when it was not. Presence or absence of database information did not have a significant effect on the outcome (odds ratio = 2.26, p = 0.354).

Correct Use of Pedestrian Pushbutton

The predictor variables were database information and APS. The interaction effect of database information and APS was not significant (p = 0.677). However, a higher percentage of participants used the pushbutton correctly (i.e., used it when present and did not look for it when absent) when the database information was provided (mean = 84%, SE = 5.8%) than when no database information was provided (mean = 52%, SE = 6.4%, odds ratio = 5.73, p = 0.005). In other words, the odds of participants using the pushbutton correctly were almost six times greater when they received the database information than when they did not. Presence or absence of APS did not have a significant effect on pushbutton behavior (odds ratio = 2.37, p = 0.145).

 

Crossing Initiation during Walk Interval

The predictor variables were database information and APS. The interaction effect was not significant (p = 0.176). However, the participants started crossing during the Walk interval (as opposed to Don鈥檛 Walk or Flashing Don鈥檛 Walk) at a higher percentage when they received the database information (mean = 70%, SE = 6.7%) than when they did not (mean = 57%, SE = 6.7%, odds ratio = 3.16, p = 0.025). In other words, the odds of participants starting to cross during the Walk interval were more than three times higher when the database information was provided than when it was not. Presence or absence of APS did not have a significant effect on the outcome (odds ratio = 0.58, p = 0.356).

Completing Crossing before the Signal Changes to Don鈥檛 Walk

The predictor variables were database information and APS. The interaction effect was significant (odds ratio = 15.56, p = 0.001), and therefore simple effects rather than main effects were examined. In the presence of APS, participants completed crossing before the signal changed to Don鈥檛 Walk at a much higher percentage when the database information was provided (mean = 64%, SE = 8.0%) than when it was not (mean = 19%, SE = 6.2%, p < 0.001). In contrast, in the absence of APS, there was no significant difference in the percentage of participants completing crossing before the signal changed to Don鈥檛 Walk between the database condition (mean = 63%, SE = 10.2%) and the no database condition (mean = 78%, SE = 7.7%, p = 0.189).

Discussion:

In general, the results suggest that the database information was more helpful for crossing tasks than for wayfinding tasks. In the database condition, participants looked for and used the pedestrian pushbutton on 73% of trials where a pushbutton was present, but did so on only 46% of trials in the no database condition, a significant difference. Presented in terms of odds ratios, the odds of looking for and using the pedestrian pushbutton were almost 6 times higher in the database condition. It did not matter whether the pushbutton was at an APS-equipped crosswalk. This is an important finding given the necessity of using the pushbutton to obtain sufficient time to cross at many modern signalized intersections. Most intersections equipped with pedestrian pushbuttons are not equipped with accessibility features such as locator tones and APS. It is essential that pedestrians with vision disabilities be alerted to the presence (and ideally, to the location) of the many silent and therefore 鈥渋nvisible鈥 pushbuttons at the crosswalks they must navigate.