3.5. Modeling Impedance of Different Transit Tasks

In this section, the field test data are examined in the light of how the environment and the placement of locations and their cues affect the blind traveler.  Kevin Lynch, in his seminal work “The Image of the City” ( 1960) used the term “legibility” to explain how certain parts of a city had features that led to a greater knowledge and awareness of feature locations and spatial interaction with other parts of the city.   This section will show that “legibility” can also be applied to various locations and transit tasks in and around a transit terminal.

3.5.1. Accessibility of Grouped Tasks and Locations

In a previous section on the field tests, each of the test destination locations were examined, and the kinds of cues they provided were discussed.   It was also noted if their spatial placement could be considered as part of the typical layout of locations in an environment, thus aiding accessibility to them, or, conversely, if their inconsistencies exacerbated the difficulty of locating them.   This section will examine these different types of location categories and then propose a model to estimate the time penalties faced by blind people in other environments.  The effect that RIAS has on these locations will also be modeled.   This will allow planners and O&M instructors to better understand the major barriers to successful independent travel for this group and how providing basic spatial information such as directional and identity cues can mitigate these problems.  
 
Previously, time penalties faced by the vision-impaired were examined, using all 30 of the subjects while they made their first attempt in the test environment.  Some of those subjects reported sufficient residual vision to potentially introduce error and noise into the models.   To increase validity, this section and the models that follow use only those 20 subjects that had no useful vision.   These subjects reported they could not see shapes or objects at all, and so the variance caused by residual eyesight can be eliminated.   There were 11 such subjects who used their regular aids first and nine such subjects who used RIAS first.  As with the travel time data for the 30 subjects, times to find the correct locations were compared to the “optimal” time based upon the familiar sighted user’s (FSU) travel time to determine the extra time it took to perform these tasks without vision.   This time penalty, caused by the lack of visual cues, can also be formulated to obtain a measure of “relative” access as compared to absolute access.   People who use wheelchairs can face physical barriers, which, in some cases, deny absolute access to a location.   Situations exist that also block some blind people from completing or even starting travel, such as difficult intersections or especially, travel in a new environment.   For blind and vision-impaired people, the barriers they face are more of a functional rather than an absolute physical nature, and with training and familiarity they might gain access.   However, the travel times are often much longer for the vision-impaired, and this can be termed a matter of relative accessibility .   They do have access to locations and opportunities, but the extra time spent searching and traveling can decrease the number and types of activities they can perform in a given time frame.   Building on Equation 1 on page 94 , a frequency variable is added.   
Equation 2  
Where:
  • f ikl = the frequency of each type of activity
     
  •  is the time or distance from   i to the desired location that offers activity k to serve a person at i with access type  l. 
     
  •  relative accessibility of   activity k from location i for person type l relative to person of type m.
     
    In its typical use, this formulation can be used to compare relative access for multiple trips to the same or similar activity.   An office worker might make three trips to the copy room each day and if that person faced travel restrictions or barriers, compared to a typical user, the time penalties would increase as the number, or frequency, of identical trips, from i to k, increased.   A modified and relaxed formulation is used here to measure trips to an activity k, from the previous location i .   For example, the subjects went to three different doors for the train boarding area, from different starting points.   By adding those trip penalties, an average time penalty for finding an unmarked gate door can be revealed.    The next section deals with measuring the relative access for various types of locations or activities.   Later, a combined relative access measure that sums up this measure for all the various activities will be given.  

    3.5.1.1. Access Problems for Specific Task s

    The first four location types examined included nine of the 20 test destinations.  Figure 3.12 shows these nine locations grouped into the four specific types.   The averages for each group are also shown, representing the frequency of each type of activity.   Later, the other 11 locations that had a wide range of non-visual cues available to the blind traveler are examined.

    Figure 3. 13   Travel Time Penalty for Four Specific Types of Tasks


    3.5.1.1.1. Track door

    Subjects started the experiment with their back at an unnamed track door, and it was explained to them that all the trains came in from behind them.  Therefore, they were aware of the spatial arrangement of all track doors being located only at the back wall of the terminal.  The doors to the train boarding area were not marked with Braille or any raised or tactile information.  Because of this, there was no way for blind subjects to identify the proper door.   Even when they found a door, they had no way to know if that was the correct door or if they should go to their left or right to continue their search.   If they did not ask for help, there was little possibility that a totally blind individual could find the proper location.   This task was also rated as the most difficult from the 26 transit tasks shown in Table 3.1 .  The difficulty of this task was also evident from the travel time data collected as subjects visited 3 different doors during the experiment.   The extra time needed to find these 3 different unlabeled doors was quite similar, and the mean time penalty was 496% more than the FSU.   The use of RIAS lowered this penalty to 208%.   These penalties could be applied to other unlabelled doors that have some order but offer no other cues to their identity.   These locations will be referred to as “Door, No Cues” or DNC.

    3.5.1.1.2. Hard Street Crossings

    Blind people are given training on crossing streets, and these locations certainly offer many non-visual cues.   However, as the experiment showed, some streets are just too dangerous for them to cross because of high-speed traffic and complicated traffic flows, such as turn lane cycles.   Subjects crossed to the mid-street transit platform (King St.) twice in the experiment, and both directions were categorized as very difficult or hard.   Some subjects refused to cross the 2 lanes to the platform on their own, and others had to wait through several cycles of the light to understand the traffic flow.   Crossing this street in both directions had a mean time penalty of 595%, while with RIAS the extra time was only 41% more than a sighted pedestrian.   These locations are referred to as “Street, Hard Difficulty” or SH.

    3.5.1.1.3. Medium Difficulty Street Crossing

    Crossing 4th Street was much different than crossing King Street.  It was a congested city block with many cars and cabs stopped at the terminal and had slow traffic.   Therefore, there were many audible cues to the traffic and turn cycle and much less danger from high-speed traffic.   Orientation and Mobility instruction and the subjects’ training are well represented in this task.  Subjects who used their regular aids and skills were able to cross this street in both directions with a mean time of 82% more than the FSU.   With RIAS, subjects were able to cross the streets with only 12% more time than the FSU.  This measure was labeled “Street, Medium Difficulty” or SM.   An easy street might be one with little traffic and stops signs.  

    3.5.1.1.4. Walking to a Street Corner

    Twice in the experiment, subjects walked to a street corner.  There are many non-visual cues to help identify a busy intersection.   Orientation and Mobility instructors also spend much time teaching these skills, and their efforts are well documented here by these results.   Subjects used dogs, their cane, and traffic noise to identify the street and its intersection.   Both of these walks were at a significant distance from the start point, but, as Figure 3.13 shows, this was not a difficult task for this group.   The mean extra time for the people using their regular aids and skills was 128% more than the FSU, and when they used RIAS they got there with only 69% more time.  This location measure is referred to as “Walk to Corner” or WC.

    For these four specific locations, it can be seen how difficult it is to find unlabeled doors and how RIAS reduces the needless search time to collect this information.   Crossing a difficult street can be such a barrier that one failed or stressful crossing may cause a trip to be abandoned.   While the blind subjects did quite well crossing the medium difficulty street, the use of RIAS in both of these street crossing tasks took away the uncertainty and stress of learning an intersection’s traffic flow, signal cycle, and other idiosyncrasies.   Without full attention to all these cues, any street crossing can lead to injury or even death.  It is no wonder that many blind people do not travel independently to new areas.   The task of walking to a corner was not too difficult for this group, but, again, RIAS helped speed up this process, especially in finding their way out of the building and to the sidewalk.  

    3.5.1.2. Location Types Based on the Availability of Non-Visual Cues . 

    The other 11 locations were not so easily categorized as to specific types of locations.   They were grouped using the “legibility” (how easy they are to understand and locate, see Lynch, 1960) , or “rationality” of their placement, as well as whether or not other cues were available to inform the blind population.   For example, it is usually a good spatial search heuristic or tactic to assume that the bathroom for one sex is near that for the other.   Ticket sales are usually in a high-traffic, central area near the entrance to a terminal or near the tracks.   Other locations, however, have no ‘standardized” or rational legibility.   They might have non-visual cues such as smells or distinctive sounds that might be heard (for example, one might hear coins at a vending machine or people using a phone or buying a ticket).   Air currents and temperature or light intensity changes can signal doorways and openings.  Other locations offer little in the way of cues to their existence.   Figure 3.14 shows the time penalty for subjects with no useful vision on their first attempt for the other 11 tasks.

    Figure 3. 14  Travel Time Penalty for Cue-based Location Tasks

    3.5.1.2.1. Random or Inconsistent Amenity Placement with No Cues
    The two hardest locations to find were also directly necessary for successful transit and transfer use.   Inconsistent placement and no cues made the bus stop and the Light Rail fare machine almost invisible to the blind person trying to use these modes.   These two locations highlight the lack of access to needed information to effectively use transit.    


    Finding a bus stop is one of the hardest tasks for blind travelers.   Indeed, subjects rated it as one of the most difficult tasks (see Table 3.1  Ratings of Transit Task Difficulty ) and research by Crandall, et al. (1996 ), Bentzen, et al. (1999 ), and Golledge & Marston (1999 ), confirms this, using other field tests.   In fact, not one of the 15 subjects in the experiment reported by Crandall and Bentzen was able to find a bus pole that was identified by tactile signs.  Bus stops can be located anywhere along the entire block face, and their signage, amenities, and cues are widely varied.  Signage can be on trees, traffic sign poles, streetlights, or a separate pole.   Stops can sometimes be identified by the location of a bench or shelter, but finding a bench does not always indicate a bus stop.   Some shelters or benches are along the curb face, while others are set back near a building line.   To make things even worse, if there are no tactile or Braille markings, even when people find a stop they have no positive feedback about which bus stops there.  These problems were clearly exposed in this experiment.   It took those who used their regular methods, including asking for help, 1970% longer than the FSU.   Those who used RIAS knew exactly where they were and identified the correct bus stop in the same time as the FSU.   This kind of positive identification can be priceless to the vision-impaired traveler and can save much time, stress, and frustration, and help increase overall access.  

    The fare machine at the Muni station was also hard to find, and it took the regular method users 1498% longer to find, while the RIAS users took only 116% longer to identify the fare machine.

    In the current experimental setup, the flower stand did not have much legibility because of the low level of activity there and the unexpectedness of this type of business being in a transit station.   There were also few cues, insofar as there was usually no one in line talking to a clerk to give any auditory cues.   It took the subjects who used their normal skills 1414% longer than the FSU, while those using RIAS found it within 251% of the standardized time.
     
    These three locations were categorized as “Inconsistent Locations and No Cues” (ILNC), and, for this type of location, the mean time penalty was 1899% longer, and the RIAS users took only 174% longer than the FSU.  
    3.5.1.2.2. Amenities with Some or Few Cues

    The bathrooms, hot dog stand, candy counter, and the outside public phones had some non-visual cues.   These types of locations had a mean penalty for the regular users of 491%.  When using RIAS, the penalty was reduced to 265% more than the FSU.   This measure is “Location, Few Cues” or LFC.

    3.5.1.2.3. Amenities with Good Cues
    The ticket window offered many cues, it almost always had a solid line of people (to bump into), and there were often voices from the people in line or at the window.  In addition, in the field test, subjects passed directly by it once before they attempted to locate it on two subsequent tasks.   The inside phones and water fountain and bathrooms were in the small waiting room just a few feet from each other.   The bathroom was the first location that was visited in this experiment.   Because of the field test order, subjects had already walked directly by the phones to get to the bathroom before they later searched for the phones.   The phones offered good cues, as there were often people talking on them or coins being inserted could be heard.   Water fountains can also offer distinctive sounds when used.   By the time they were to locate the water fountain, they had already been in the immediate area twice; for the bathroom and for the phones.   Search and exploration of the environment allowed some subjects to find these locations, or gain valuable cues, while searching for other locations.  

    For these four locations that offered good non-visual cues, the mean time penalty was 260%, and, for the RIAS users, 217% more than the FSU.   This measure is called “Location, Good Cues” or LGC.     Because the fixed order of the location search tasks required multiple exposures in these areas, these four locations had too many confounds to be valid for modeling purposes.   They were included here in the explanation but will not be discussed in the next sections on location difficulty coefficients or in the models. The order of the search tasks caused some of these locations to be easier to find than would be the case if a person searched for them only when needed.   These results are a confound of the station layout and experiment task order and should not be interpreted to apply to other locations of the same type.   Even though an area like the ticket window, with its distinctive sounds and lines of people, posed less difficulty than most other locations, it is a vital and necessity part of each traveler’s transit experience.   These high traffic demand and necessary amenities should be given as many cues as possible. 

    3.5.1.3. Summary of Location Tasks

    Unlike the person with physical mobility impairments, such as severe arthritis, a bad hip, chronic fatigue, or a weak heart, there is no consistent time penalty that can be measured relating to the travel time of blind people.  These data show that the problems that cause a blind person to travel with less efficiency in an environment are not necessarily some inherent disadvantage caused by the lack of vision.  Inconsistent locations with no cues and doors with no labels cause large time penalties and stressful travel, while locations with more environmental cues are much easier to find.   It appears that it is often the lack of directional and location identity cues that cause the inefficient travel behavior (longer travel times) exhibited by many blind travelers. 

    3.5.2. Coefficients of Location Difficulty and Successful Mitigation

    Time penalties increase as the number and types of trips increase.  A more active traveler, who faces barriers to efficient travel, has more cumulative penalties than an inactive person.  By summing up Equation 2 on page 170 , a formulation can be presented that compares two types of users, with different access mode criteria, over a wide range of activities.   This formulation can be used to compare the daily, weekly, or longer variation in travel time for different groups.   The cumulative relative access measure thus allows for examination of how time penalties combine, depending on the choice of activities, to restrict access due to time constraints.  

    Equation 3

    This equation is the same as Equation 2, except the time penalties are added together.  Using this formulation, the access mode type can be varied to examine the overall time penalties or relative access measures.   This formulation is modified so that starting location i is relaxed to mean any location i for a trip to activity or location k.   For example, the mean time penalty for trips to the doors is added to the penalty for crossing the hard street, and all the other types of locations, to produce the total time penalty of the 20 destinations in this experiment.    Table 3.8 shows five different ways to judge the time penalties faced by people with vision restrictions.

    Table 3. 8   Impedance Coefficients for Various Locations

     

    Specific Tasks and Locations

    General Locations

    Coefficients of Difficulty for Transit Tasks

    Door

    No Cues

    Hard

    Street

    Med.

    Street

    Corner

    Walk

    Location

    No Cues

    Location

    Few Cues

    Variable Name

    DNC

    SH

    SM

    WC

    ILNC

    LFC

    Blind, Regular Method /

    Sighted Baseline

    6.0

    6.9

    1.8

    2.3

    20.0

    5.8

    Blind, with RIAS /

    Sighted Baseline

    3.1

    1.4

    1.1

    1.7

    2.7

    3.7

    Blind, Regular Method –

    Blind, with RIAS

    2.9

    5.5

    0.7

    0.6

    17.2

    2.2

    Blind, Regular Method /

    Blind, with RIAS

    1.9

    4.9

    1.6

    1.3

    7.3

    1.6

    % Time Saved with RIAS versus Regular Method

    48%

    80%

    38%

    26%

    86%

    37%
     
    A short discussion of Table 3.8 follows for the five rows of difficulty coefficients.   The location variables with the highest degree of difficulty were (in decreasing order) ILNC, SH, DNC and LFC.   The coefficients ranged from 20.0 to 5.8.   These types of locations can be so inconsistent in placement, legibility, safety, and availability of cues that there is no effective way to be trained to find them.  The less difficult location variables were the WC and SM.   These last two locations require skills that are well learned with O&M instruction, training, and practice.   These “less difficult” tasks still had penalty coefficients from 2.3 to 1.8.
     
    When using the RIAS, the difficulty coefficients drop to a range of 3.7 to 1.1.  Using RIAS lowered the difficulty coefficients of all six location variables.   The biggest savings were for the location variable ILNC, where the penalty was lowered by 17.2 (from 20.0 to 2.7).   The next three locations most improved by RIAS were SH, DNC, and LFC, with a savings range from 5.5 to 2.2.   Even the lowest savings, WC and SM, were 0.7 and 0.6 times the FSU respectively.

    The same pattern exists when one computes the time penalty of regular methods over that for RIAS.  ILNC, SH, and DNC were still the most difficult locations, when compared to RIAS, with a range of 7.3 to 1.9, while the less difficult tasks were SM, LFC, and WC with a difficulty rating of 1.6 to 1.3 more than when using RIAS.

    It is important to realize how much time could be saved with the addition of directional and identity cues in an environment that is lacking cues for the blind traveler.  Using RIAS saved people searching for ILNC locations 86% of the regular method time, and it saved 80% of the time it took to normally cross a difficult street (SH).  For location types DNC, SM, and LFC, the savings ranged from 48% to 37%.   Even the lowest savings were notable, with the WC task saving 26% of the time that it took people to find these locations using their regular aids and travel skills.

    3.5.3. Modeling Transit Task Difficulty and Mitigation

    Using the above location time penalty coefficients, models can be produced that will assist people interested in navigation without sight, especially planners and O&M instructors, to apply these findings to other environments.   Producing a linear model of both experimental conditions and also of the time saved between the conditions can identify more completely which types of tasks present the most resistance to efficient travel.   Three linear models are presented that can be used to estimate the total travel time required for a blind traveler, based on the time for a sighted and familiar user.  Prudent application of these models would allow a better understanding of the difficulties that people without sight might face in a new environment, without the need to collect data from a group of blind users first.   Architects and design professionals, especially transit planners, could test their designs before they are built in order to ensure the best compliance with ADA mandates. These models could help planners know where to concentrate their mitigation efforts and add to the body of knowledge about barriers to accessibility in urban environments.   As the models show, it is the environment, placement of destinations, and lack of cues that helps create the penalty to navigation without sight much more than the inherent lack of vision itself.   A better designed and equipped environment would go a long way to ensure that this group could use the facilities with independence, efficiency, and dignity and would make the travel experience less stressful and provide a higher degree of personal safety.   Simply stated, the model takes the time penalty coefficients for each of the six location or activity types and multiplies them by the time it takes for a sighted traveler to complete the tasks.   When those numbers are summed, it reveals the total time penalty.

    The model is based on a linear model with the equation:
    Y= e +B1X1 +B2X 2 +B 3 X3  +B 4 X4 +B5X5 +B6X 6
    where:

    The first model computes the extra time that blind travelers expend in different locations, compared to a baseline (FSU) time.   (Y) is the predicted value of the time it would take for the blind to complete the tasks.
    Equation 4
    Model 1: Y = e +(6.0)DNC+(6.9)SH +(1.8)SM   +(2.3)WC+(20.0)ILNC+(5.8)LFC

    It can be seen that  “inconsistent locations, no cues” (ILNC) has a time penalty of 20, and crossing the difficult street (SH) has a penalty of 6.9.   In contrast, a location with little or few cues (LFC) has a penalty of only 5.8 and crossing the medium difficulty street has a penalty of only 1.8.  This type of model can be used in several ways.  It is easy to see that adding a few cues at certain locations (changing a location from ILNC to LFC) would reduce the overall time penalty for a trip.   In addition, re-routing the trip could also reduce overall penalties.   In this example, it would be faster to cross two medium difficulty streets (SM) than to cross one street with hard difficulty (SH).   The goal of increasing access can be met by designing spaces to reduce the cumulative penalty (Y).
      
    The second model allows for computation of the reduced time penalty in an environment if directional and identity cues were available, as when using RIAS.  (Y) is the predicted value of the time it would take for the blind to complete the tasks when using RIAS, as compared to a sighted user.
    Equation 5
    Model 2: Y = e +(3.1)DNC+(1.4)SH +(1.1)SM   +(1.7)WC+(2.7)ILNC+(3.7)LFC

    In this model, it is clear that the time penalty between hard (SH) and medium difficulty (SM) streets has become quite similar, unlike in Model 1, and other penalties are smaller as well.   This would produce a much lower total penalty (Y) than Model 1, for the same route and activities.  
     
    To determine how much time could be saved when a blind person has access to additional auditory cues, a third model shows the effect on environments’ “legibility” and ease of use when a system like RIAS is installed.   (Y) is the predicted value of the time saved when using RIAS to the time of the blind using their regular methods.   In this model, X1 – X6 are the walk and search times of the regular blind user.  
    Equation 6
    Model 3: Y= e +(48%)DNC+(80%)SH +(38%)SM  +(26%)WC+(86%)ILNC+(37%)LFC

    Using the mean travel time of blind travelers, this model can estimate the savings when using accessible cues, such as RIAS.   For example, it shows that RIAS might save 48% of the time to find unlabeled doors (DNC), 80% to cross hard streets (SH), and 86% at those locations that are inconsistent and have no cues (ILNC).

    3.5.4. Section Summary

    There is no consistent restriction or time penalty that can be assigned to the search times for blind travelers.   These data and the models should allow planners to consider which locations demand attention in order to help mitigate barriers to access.   Spatial knowledge acquisition, especially for people who are blind, can be increased with proper attention to the consistent location of amenities.   Accessibility for the blind can also be increased by giving more attention to providing cues to these locations, including the use of identity and directional cues as provided by RIAS.   The continued existence and acceptance of such high penalties and barriers to independent travel should be robustly questioned and examined by anyone concerned about providing access to urban opportunities and an equitable society for all people.

    BACK TO OVERVIEW BACK TO TABLE OF CONTENTS NEXT SECTION