3. Specific Transit Tasks and Locations That Restrict Travel

Hypothesis 1: Experiment data will show that, for those with limited vision, specific locations and tasks cause difficulty when using transit. The use of auditory signage will mitigate much of the difficulty. Previous research gave strong indications that specific transit tasks are identified by blind travelers as causing time and effort constraints on their travel and can also lead to a person not making a trip (Golledge et al., 1995; Golledge, Costanzo, & Marston, 1996; Golledge & Marston, 1999; Golledge, Marston, & Costanzo, 1996; Golledge et al., 1997; Golledge et al., 1998; Marston & Golledge, 1998a, 1998b; Marston & Golledge, 2000; Marston et al., 1997) .

The first section describes the field test experiment where vision-impaired subjects performed various transit tasks with their regular method of guidance and also with RIAS. This research investigates how specific locations and tasks are functional barriers to efficient navigation, and how the use of RIAS might mitigate many of these barriers, providing a much higher level of informed and effective transit use. Results of the timed trials are discussed and analyzed. Later in this chapter specific location difficulties, as reported by subjects through responses to questions, will be examined.

3.1. Caltrain Field Test

3.1.1. Procedures

The complete instructions given to the subjects are listed in the questionnaire in APPENDIX 4 (under the field test section.) In addition to transferring from one mode to another, the experiment was made more realistic by requiring the subjects to find different amenities along the route, such as ticket windows, bathrooms, public phones, etc. Subjects were told which locations to search for or which direction to go to find a street corner. They were allowed to ask others (but not the researcher) for verbal assistance only.

Task 1: Subjects were walked, in a disorienting fashion, to the doors leading to track 7 at the Caltrain station. They were told to imagine they had just disembarked from the train and had entered the station. With their back to the train track door, the researcher took one of their hands and drew an upside down “T” on their open palm to show the shape of the interior. It was explained that the train tracks were behind them, opened into a long hallway, and that the main hall to the exit was in front and toward the left. Subjects were told that the terminal amenities were either located in the main hall or nearby on the opposite wall. This was the only spatial information they were given about the site layout. Their task was to first find the proper bathroom and then find where to buy a candy bar. From there, they were to walk out the station’s main entrance, turn right, and go to the corner. After listening for at least one cycle of the traffic signal, they were to cross King Street to the median Muni platform. They were required to tell the researcher when they wanted to cross, so that the researcher could monitor their safety. Once at the platform, they were to find a fare machine, which sold tickets for the Muni Light Rail station.

Task 2: Subjects started at the mid-street platform corner by the Muni station fare machine. They informed the researcher when they wanted to cross and then were to cross from the platform to the Caltrain side of King Street. From there, they were to walk back to the Caltrain station and find the ticket window. Subjects then were to search for the flower stand, and then walk to the bank of pay phones inside the station. From there, they were told to find the door for gate 2.

Task 3: Because of construction barriers, subjects were led from gate 2 and proceeded out the main entrance of the station where they were to turn left toward Townsend Street and left again down Townsend to the cabstand. Walking independently, they were told to choose any route in order to locate the water fountain, then locate the ticket window, and finally, locate the door for gate 11.

Task 4: Subjects left gate 11 and were told to return to the first corner that they had visited, the one across from the Muni station. However, here they were to cross the street (4^th St) in front of the station. Again, subjects notified the researcher before they attempted to cross 4^th street. Once across the street, they were to turn left and find a pay phone further down the street. After finding the pay phone, they were to locate the bus shelter for the Muni #15 bus line.

Task 5: Subjects started at the corner of 4^th and King. Here they independently crossed the street toward the Caltrain station, and then the researcher took them to the ticket window in the station. From the window, they independently searched for the concession stand that sold hot dogs and then searched for and walked to the door for gate 3.

For each of these five transfer tasks, data were collected on the time it took to complete each leg of the task, the number and types of errors made, and the number of times they asked for help from others.

All times recorded were in seconds. A maximum of four minutes (240 seconds) was allowed for each sub-task. For this task, 15 subjects used their regular skills first (NRIAS) for all five sub-tasks and then later repeated the same tasks using the RIAS. Fifteen subjects used the RIAS first and 10 people repeated the task later using their regular skills. The t-tests statistics were calculated for analysis of mean times between the two conditions, NRIAS 1^stversus RIAS 2^nd and RIAS 1^stversus NRIAS 2^nd. The t-test statistics were also calculated on the difference between the 30 RIAS scores and the 25 NRIAS scores, regardless of the order of the condition. The results are presented next for each sub-task of these transfer tasks.

Figure 3. 1 Transfer Task 1 Path of Travel

3.1.2. Transfer Task 1: Track 7 To Muni Fare Box

All Times in seconds (s)

NRIAS = No RIAS RIAS = Using RIAS

			Condition 1		Condition 2
Task	From	To	NRIAS 1^st	RIAS 2^nd	NRIAS 2^nd	RIAS 1^st
1-A	Track #7	Bathroom	142s	60s	92s	85s

The difference in times when using RIAS after the regular method was highly significant (p<.0005). There was no significant difference when using RIAS first and then the regular method (p<.4). Overall, the difference between the two conditions was highly significant (p<.003).

Ten subjects asked for help from others 13 times when using their regular method. No one using RIAS asked for help.

			Condition 1		Condition 2
Task	From	To	NRIAS 1^st	RIAS 2^nd	NRIAS 2^nd	RIAS 1^st
1-B	Bathroom	Candy	134s	86s	81s	112s

The difference in times when using RIAS after the regular method was highly significant (p<.008). There was no significant difference when using RIAS first and then the regular method (p<.08). Overall, the difference between the two conditions was not significant (p<.23).

Eight subjects asked for help from others nine times when using their regular method. No one using RIAS asked for help

			Condition 1		Condition 2
Task	From	To	NRIAS 1^st	RIAS 2^nd	NRIAS 2^nd	RIAS 1^st
1-C	Candy	Corner	134s	102s	131s	114s

The difference in times when using RIAS after the regular method was highly significant (p<.0006). There was no significant difference when using RIAS first and then the regular method (p<.24). Overall, the difference between the two conditions was not significant (p<.06).

Vision-impaired people are quite used to using traffic sounds and the cane or dog to find a street corner, and no one asked for help on this task.

			Condition 1		Condition 2
Task	From	To	NRIAS 1^st	RIAS 2^nd	NRIAS 2^nd	RIAS 1^st
1-D	Corner	Corner	48s	12s	42s	13s

Knowing when to cross a busy street can be a difficult task, depending on intersection types, turn lanes, and traffic flow. RIAS gives a distant and definite “WALK” or “WAIT” signal, and this advantage is clearly shown at this crossing. The difference in times when using RIAS instead of the normal method was highly significant in both condition orders and also overall (p<.006, p<.01, and p<.0001 respectively). Eight subjects out of 25 using their regular method made a total of 15 unsafe attempts to cross the street. Nobody using RIAS made any unsafe attempts, again showing the benefits in terms of safety for the user with RIAS. In addition, one subject completely missed the opposite corner when using the regular method of travel.

			Condition 1		Condition 2
Task	From	To	NRIAS 1^st	RIAS 2^nd	NRIAS 2^nd	RIAS 1^st
1-E	Corner	Fare Box	140s	15s	30s	21s

The difference in times when using RIAS after the regular method was highly significant (p<.00006). The fare machine was in a very inconspicuous spot, and, without RIAS, many people missed it completely. Those that used RIAS first appeared to learn this location well and were able to find it much easier the second time after having used the RIAS. There was no significant difference when using RIAS first and then the regular method (p<.14). Overall, the difference between the two conditions was highly significant (p<.00001). One subject asked for outside help.

	Condition 1		Condition 2
	NRIAS 1^st	RIAS 2^nd	NRIAS 2^nd	RIAS 1^st
Task #1 Total	596s	277s	374s	345s

The total of the five sub-tasks that make up Transfer Task 1 shows how much better people traveled when using RIAS. Once having used the system, their spatial knowledge appears to increase so that, on their second attempt using their regular method, the results, although quicker in the RIAS condition, show no significant difference (p<.34). When using RIAS 2^nd after their regular method, the results are highly significant (p<.00001). The results are highly significant over all the trials for the two conditions (p<.0004). There was no significant difference when using RIAS first or second, showing that the initial trial with the regular method did not help them learn the route (p<.11). For the 25 subjects who attempted the five sub-tasks with their regular method, there were a total of 21 tasks that they could not finish and were “timed out.” The group of 30 subjects attempting the same five sub-tasks with RIAS only had two that were “timed out.” The data for the five transfer tasks are shown in APPENDICES 5 through 9.

3.1.2.1. Time Penalties and Accessibility

To identify how specific locations and tasks affect blind navigation, the travel times in the two conditions to the FSU (familiar sighted user) were compared (see Section 1.6.6 , Sighted Subjects for Baseline ). The excess travel time in the two conditions, compared to the FSU, is expressed as a percentage of the baseline time. This standardized the effect of different distances between the task locations. This method shows the time penalty faced by people with vision loss, and how that penalty is not consistent, but varies according to the characteristics of various locations, including the non-visual cues available. This method allows for better understanding of which locations present the biggest barriers to successful and independent travel. Examining the time penalty also shows how the use of RIAS affects the time required to perform these tasks. A few caveats to better understand these data are:

This experiment was conducted in a busy area with various obstacles, such as crowds, that varied during the experiment.

Different locations offered various degrees of non-visual cues that affect the data, and, as the experiment went on, some locations had been discovered earlier.

These times represent not just finding the location but navigating the path from the previous location as well.

Although all subjects were allowed to ask others for help, no users of RIAS asked for help and this sometimes affected their performance to a degree.

A few locations had signals that were partially blocked or obscured at certain locations.

Since there was a 4-minute maximum allowed for each sub-task, walks that took longer for the FSU did not have excess time penalties as high as those for shorter walks.

These findings for each location will be discussed and explanatory comments from the researcher’s information will be offered later (see Section 3.5 , Modeling Impedance of Different Transit Tasks ). The figures in this section on the five transfer tasks use data from the 30 subjects on their first attempt only. This removes any effect of learning from a second trial and increases the validity. Therefore, it is a between subjects test with fifteen subjects in each condition.

3.1.2.2. Time Penalty Formulation

One way to measure how access is restricted for certain groups or individuals is to compare the effort of travel (such as time or distance) for that group or individual to a user with less restricted travel means. For example, the time it takes to commute by bus can be compared to the use of a private car to show if there is extra time spent by using the transit mode. In this section, the travel time of legally blind people is compared to a sighted person, to determine the excess time, or penalty, for travel without vision. If a person who was blind took 10 minutes to walk from the entrance of a train station to the proper boarding gate and a sighted person took only two minutes, there would be an extra time penalty of eight minutes for the trip without sight. A ratio accessibility measure can be formulated to show the excess time or distance required by people who have restrictions on their travel. This can be shown as the additional extra time compared to the sighted traveler. For this example, it would be expressed as ((10/2) –1), or 4 times that required with sight. Relative accessibility (in this case the extra travel time --see Church & Marston, (in press) , can be formulated as:

Equation 1

where:

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 .

Person of type m is the sighted walker, and in the above example, 4 is the relative accessibility measure, or time penalty, when comparing a trip from the station door to a boarding area for the two travelers. In the following sections, a person of type l is used to represent the blind person in either of the two test conditions, using regular navigation skills (NRIAS ) or using RIAS. With this formulation, a relative access score of 0.0 would represent a location that could be reached by the sighted and the blind traveler in the same amount of time. A score higher than 0.0 would represent an excess time penalty caused by lack of vision. Figure 3.2 compares the excess travel time required for blind people with and without the use of RIAS.
Figure 3. 2 Excess Time using Regular Methods and RIAS - Task 1

3.1.2.2.1. Fare Box

In transfer task 1, the hardest location to find was the fare box at the Muni rail station. The entrance ramp to the station was itself hard to locate, and, since there were no turnstiles to signal the paid area, many people missed the fare box. It was placed in a position that did not correspond with typical “environmental grammar,” meaning a common and consistent location. This location was categorized as an inconsistent or random transit amenity location with no cues. The 15 subjects using their regular method of travel took 1168% longer than the familiar sighted user (FSU). A RIAS signal allowed easy access for the other 15 subjects who only took 87% more time than the
FSU.

Using the first trial data only, t-test statistics for the two conditions, regular method and RIAS, showed a significant difference (P<.00002). After subjects that reported they could see shapes or objects up close were eliminated, subjects who had no vision at all had t-tests that showed a significant difference (P<.00002). The subjects with no useful vision are discussed later (see Section 3.5.1 , Accessibility of Grouped Tasks and Locations ).

3.1.2.2.2. Bathroom

The bathroom was located in a waiting room off the main terminal area. Once the waiting room is located, there are many obstacles of chairs, people, and pillars. The bathroom doors had the standard round (F) or triangle (M) tactile information. This location was categorized as an amenity with few cues. Subjects using their regular method took 575% longer than the FSU. The RIAS subjects took 304% longer than the FSU, partially due to the various obstacles in the waiting room.

The t-test statistics for the two conditions showed a significant difference (P<.01). For those subjects with no vision, t-tests showed a significant difference (P<.03).

3.1.2.2.3. Hard Street Crossing #1

Blind travelers use auditory cues from traffic to align themselves for a street crossing and also to understand the flow of traffic. Turn lanes and high-speed traffic confound these problems. This crossing of King Street was quite difficult for many blind people. King Street is a high-speed arterial road, and the nearest stop light from this one is two long blocks away so traffic moves at a high speed. In addition, most of the cars on 4^th Street turn right onto King, so there is almost a constant flow of traffic, except for the short walk cycle (see Figure 2.6 on page 59 ). This location was categorized as a hard street crossing. Regular method subjects took 377% longer than the FSU, and, with RIAS, users took only 31% longer.

The t-test statistics for the two conditions showed a significant difference (P<.004). For those subjects with no vision, t-tests showed a significant difference (P<.009).

3.1.2.2.4. Candy Counter

This amenity was located in the main entrance hall of the terminal. There were often people around and the voices of the counter clerks were audible. The counter was “L” shaped and quite long and this arrangement provided much room for errors by the subjects. Usually, subjects found a part of a counter and then asked others if candy was sold there. The smells of popcorn and candy did allow some users to locate the area when close. Only after getting a verbal response from the clerk did they know their location. This amenity was categorized as one with few cues. The regular method subjects took 332% longer than the FSU, and, with RIAS, this time was cut to 262%.

The t-test statistics for the two conditions showed no significant difference. For those subjects with no vision, t-tests also showed no significant difference.

3.1.2.2.5. Walk to Corner #1

From the candy counter, subjects were to walk out the main door, turned right, and walk to the corner. Both long cane and dog users are well trained in following curbs and using auditory cues to find street corners. The regular method subjects took 147% longer, and the RIAS subjects took 111% longer, than the FSU.

The t-test statistics for the two conditions showed no significant difference. For those subjects with no vision, t-tests showed no significant difference.

Figure 3. 3 Transfer Task 2 Path of Travel

3.1.3. Transfer Task 2: Muni Corner to Track 3

For this task, 15 subjects used their regular skills first for all five sub-tasks, and then they later repeated the same tasks using RIAS. Fifteen subjects used the RIAS first and 10 people repeated the task later using their regular skills. The t-tests statistics were calculated for analysis of times between the two conditions, NRIAS 1^stversus RIAS 2^nd and RIAS 1^stversus NRIAS 2^nd . The t-test statistics were also calculated for the 30 RIAS scores and the 25 NRIAS scores, regardless of the order of the condition.

			Condition 1		Condition 2
Task	From	To	NRIAS 1^st	RIAS 2^nd	NRIAS 2^nd	RIAS 1^st
2-A	Corner	Corner	72s	13s	74s	15s

Because of the turn lanes and traffic flow at this crossing, the effects of the RIAS were highly significant. Without RIAS there was much hesitation and many mistakes. The results for the NRIAS 1^st condition were p<.006, for RIAS 1^st p<.002, and for 30 RIAS subjects and 25 NRIAS subjects, regardless of order, the results were also highly significant (p<.00004). There is no “learning” effect over two attempts at a dangerous crossing like this one.

Thirteen subjects out of 25 without RIAS made a total of 20 unsafe attempts to cross the street. Twelve of the 25 subjects using their regular method missed the corner, another dangerous situation when traveling without vision. One person out of the 30 using RIAS missed the corner. Two subjects not using the system refused to even attempt the street crossing. The times and errors show that, when using RIAS, there was little hesitation, and that safety was vastly increased.

			Condition 1		Condition 2
Task	From	To	NRIAS 1^st	RIAS 2^nd	NRIAS 2^nd	RIAS 1^st
2-B	Corner	Ticket Win	128s	100s	115s	107s

The difference in times when using RIAS after the regular method was highly significant (p<.017). There was no significant difference when using RIAS first and then the regular method (p<.47). Overall, the difference between the two conditions was not significant (p<.085).

Four subjects asked for help from others a total of four times when using their regular method. No one using RIAS asked for help

			Condition 1		Condition 2
Task	From	To	NRIAS 1^st	RIAS 2^nd	NRIAS 2^nd	RIAS 1^st
2-C	Ticket Win	Flowers	93s	15s	45s	21s

The difference in times when using RIAS after the regular method was highly significant (p<.0006). There was no significant difference when using RIAS first and then the regular method (p<.069). Overall, the difference between the two conditions was highly significant (p<.00006).

Seven out of 25 subjects asked for help from others a total of 11 times when using their regular method. No one using RIAS asked for help.

			Condition 1		Condition 2
Task	From	To	NRIAS 1^st	RIAS 2^nd	NRIAS 2^nd	RIAS 1^st
2-D	Flowers	Phone	109s	101s	80s	109s

No significant difference was found for this task. Seven out of 25 subjects asked for help from others a total of eight times when using their regular method. No one using RIAS asked for help. In this location, the light beam did not extend complexly to the adjacent wall, where most of the subjects walked as they shorelined along the building wall, causing trouble in picking up the signal easily.

			Condition 1		Condition 2
Task	From	To	NRIAS 1^st	RIAS 2^nd	NRIAS 2^nd	RIAS 1^st
2-E	Phone	Track #2	172s	86s	107s	85s

The difference in times when using RIAS after the regular method was highly significant (p<.0002). There was no significant difference when using RIAS first and then the regular method (p<.14). Overall, the difference between the two conditions was highly significant (p<.0001).

Seven out of 25 subjects asked for help from others a total of 12 times when using their regular method. No one using RIAS asked for help. Four people reported they were “not sure” they were at the correct track when using their regular method. There was no Braille signage at these doors, and, if people were not around to ask, blind people have no confirmation of the correct location.

	Condition 1		Condition 2
	NRIAS 1^st	RIAS 2^nd	NRIAS 2^nd	RIAS 1^st
Task #2 Total	574s	315s	421s	388s

The total of the five sub-tasks that make up Transfer Task 2 show how much better people traveled when using RIAS. Once having used the system, their spatial knowledge appeared to increase so that, on their second attempt using their regular method, the results, although quicker in the RIAS condition, are not significant (p<.14). When using their regular method of travel first and then using RIAS, the results were highly significant (p<.00001). The difference between the 30 trials with RIAS and the 25 trials without RIAS was also highly significant (p<.0006). For the 25 subjects who attempted the 5 sub-tasks with their regular method, there were a total of 17 tasks that they could not finish and were “timed out.” The 30 subjects attempting the same five sub-tasks with RIAS only had three that were “timed out.”

The order of the RIAS condition was shown not to be significant in this task. Comparing RIAS 1^st to RIAS 2^nd gives a value of (p<.35). This shows that the improvement in performance when using the system is not due to the learning effect of a second trial.

3.1.3.1. Time Penalties and Accessibility

Figure 3. 4 Excess Time using Regular Methods and RIAS - Task 2

3.1.3.1.1. Flower Stand

The hardest location to find in the second transfer test was the flower concession, which was located in the main area of the terminal. Subjects could not find the counter, because flowerpots on the floor in front of the counter seemed to obscure the counter. Subjects without the RIAS took 1063% extra time over the FSU to find the counter. First attempts by those with RIAS took only 162% more time than the FSU. Once they picked up the signal and got positive identification, they knew to push forward through the scattered flowerpots to the desired counter. Because of the blockage, this location was categorized as another random or inconsistent amenity location with no cues.

The t-test statistics for the two conditions showed a significant difference (P<.0007). For those subjects with no vision, t-tests showed a significant difference (P<.0006).

3.1.3.1.2. Hard Street Crossing #2

The second crossing of King Street was in the opposite direction than the first crossing. This direction, toward the Caltrain terminal, was even harder, because the traffic turning right from 4^th St. was across the street and harder to hear and comprehend. This problem and the high speeds kept some people from even attempting the crossing, and others took time to listen to 2 or 3 signal cycles before crossing. The regular method subjects took 504% longer than the baseline FSU. The RIAS gave a direct beam and positive identification of the WALK signal and users with RIAS crossed this street only 22% longer than the baseline FSU.

The t-test statistics for the two conditions showed a significant difference (P<.005). For those subjects with no vision, t-tests showed a significant difference (P<.007).

3.1.3.1.3. Train Track #2

The entrance doors to the boarding area for outbound trains had no Braille or tactile information to identify the proper door. It took 454% longer than the FSU to walk to, locate, and identify the proper door for those using their regular method. The RIAS gave a direction beam and positive identification of the track number, and those who used the system were only 176% longer than the FSU.

The t-test statistics for the two conditions showed a significant difference (P<.00004). For those subjects with no vision, t-tests showed a significant difference (P< 0001).

3.1.3.1.4. Inside Phone

The phone bank in the terminal was located in the waiting room near the bathrooms. Subjects had passed them on their first trip to the waiting room. There are non-visual cues available if people are on the phone talking or coins are heard. The RIAS signal at the waiting room was partially blocked by building columns, and inside the room there were problems with many obstacles and crowds. The excess times were quite similar at this location, with the regular method subjects taking 289% longer and the RIAS users taking 290% longer than the FSU. Since the subjects had been past this location previously and there are other auditory cues, this amenity location was categorized as one with good cues. The t-test statistics for the two conditions showed no significant difference. For those subjects with no vision, t-tests also showed no significant difference.

3.1.3.1.5. Ticket Window 1^st time

The ticket window is rationally located directly inside the terminal. For this test, the subjects walked back from the corner, retracing along the curb to the building entrance. Subjects had passed this area previously, there were usually lines of people queued up, and there were theatre-type stanchions with ribbon-tape designating the waiting area. There were also voices from the window agent and from people in line. Subjects using their own aids took 125% longer, and those with RIAS took 88% longer than the FSU. This location was categorized as an amenity with good cues. There was no significant difference for the first attempt in either the full all subjects and the no vision subjects’ data.

Figure 3. 5 Transfer Task 3 Path of Travel

3.1.4. Transfer Task 3: Taxi Stand to Track 11

For this task, 15 subjects used their regular skills first for all three tasks and then repeated the same tasks later using the RIAS. Fifteen subjects used the RIAS for their first and only trial. They did not repeat the experiment with their regular method.

			Condition 1		Condition 2
Task	From	To	NRIAS 1^st	RIAS 2^nd	RIAS 1^st
3-A	Taxi stand	Water	174s	141s	128s

The water fountain was quite distant from the cab stand and was difficult to locate in either condition for some subjects The difference in times when using RIAS after the regular method was significant (p<.01). Overall, the difference between the two conditions was significant (p<.045).

Without RIAS , seven out of 15 subjects asked for outside help a total of 10 times. In this trial, subjects were told to find the water fountain, with no mention of any specific path. They had previously been led out the front entrance and around to the side of the terminal, never using the side door (Townsend St). Six of the 15 NRIAS subjects (40%) made a shortcut through the side door of the terminal. When using the RIAS , 29 out of 30 subjects (97%) made a shortcut through the side door. This was quite revealing, because many blind people have trouble making shortcuts in an unknown space. Some of the subjects had some residual vision, but, while using the RIAS, even the totally blind were able to understand the spatial layout and find the side door entrance that they had never used. As shown in Figure 3.5 , the side doors had identifying RIAS transmitters, and it appears that although they did not use those doors previously, they must have learned and stored that knowledge on the previous tasks.

			Condition 1		Condition 2
Task	From	To	NRIAS 1^st	RIAS 2^nd	RIAS 1^st
3-B	Water	Ticket Win	81s	51s	65s

The results in the NRIAS 1^st condition were significant (p<.02). In this sub-task, the results between the two conditions were not significant (p<.09). Without RIAS, three out of 15 subjects asked for outside help. This was their second trip to the station ticket window, and, by this time, it appeared that the subjects were learning where it was located.

			Condition 1		Condition 2
Task	From	To	NRIAS 1^st	RIAS 2^nd	RIAS 1^st
3-C	Ticket Win	Track #11	178s	79s	99s

Track gate doors are not marked with Braille, and this door was at the far end of the terminal where often there were no people to ask for help. People with vision restrictions often rely on asking for help from others, but there are many situations where few, if any, people are available for assistance. This was certainly the case in this task. Six of fifteen people without RIAS could not find the door in the four minutes allowed. All 30 subjects using the RIAS found the correct track door. The results for those who used the system second were highly significant (p<.00001), and the overall results wee also highly significant (p<.000002).

In addition, over half (8/15) of the subjects not using the RIAS asked for outside help a total of nine times. Three of the regular method users also reported they were not sure if they were at the proper door, although they were actually there.

	Condition 1		Condition 2
	NRIAS 1^st	RIAS 2^nd	RIAS 1^st
Task #3 Total	433s	272s	302s

For the entire trip from the cabstand to track 11, the results were highly significant for both the NRIAS 1^st – RIAS 2^nd condition, and the overall average (p<.00002 and p<.002 respectively). For the 15 subjects who attempted the three sub-tasks with their regular method, there were a total of 11 tasks that they could not finish and were “timed out.” The 30 subjects attempting the same three sub-tasks with RIAS only had five that were “timed out.”

The order of the RIAS condition was shown not to be important in this task. People performed just as well if they used the system first or second. There was no significant difference based on order of use. The t-test showed that (p<.3).

3.1.4.1. Time Penalties and Accessibility

Figure 3. 6 Excess Time using Regular Methods and RIAS - Task 3

Train Track #11

This door was located at the far end of the terminal where subjects had not yet traveled. That area of the station was much less crowded and offered fewer people to ask for help. Those subjects using their regular navigation aids took 334% longer than the FSU. With RIAS, subjects could “see” the door numbers as they walked down the hall and were able to keep going until they found the proper door. Many subjects without the system were not aware of how many gate doors were in the station and often stopped short. Those using the RIAS were able to locate the proper door within a time period that was 141% longer than the baseline.

The t-test statistics for the 2 conditions showed a significant difference (P<.0006). For those subjects with no vision, t-tests showed a significant difference (P<.00002).

3.1.4.1.1. Ticket Window 2^nd time

Regular method subjects took 307% longer, and RIAS subjects took 226% longer than the FSU baseline. No significant difference was found for this task on the first trial.

3.1.4.1.2. Water Fountain

The water fountain was also located in the waiting room. Subjects had been there twice before this sub-task. They could have found it or heard cues on those trials. This location was classified as an amenity with good cues. To get to the start location for this path, subjects were guided out the main entrance and around the building. There was a shortcut from the start point to the water fountain through side doors of the terminal, which the subjects had not used before. The shortcuts are discussed later (see Section 4.6.1 . Spatial Knowledge Revealed by Navigation and Wayfinding Tasks ). For the excess time comparison, and because the subjects had no previous knowledge of the side doors, a comparison was made between these times and those of the FSU taking the path subjects learned on their guided walk. Subjects using their own aids took 117% longer and RIAS subjects took 73% longer than the baseline. When subjects’ times are compared to the FSU taking the shortcut, the regular aids subjects took 185% longer and the RIAS subjects took 126% longer. The two conditions were almost significant (P<.052) for the full group and not significant for those with no vision.

3.1.5. Transfer Task 4: Track 11 to Bus Shelter Line #15

For this task, 15 subjects used their regular skills first for all three tasks and then repeated the same tasks later using the RIAS. Fifteen subjects used the RIAS for their first and only trial. They did not repeat the experiment with their regular method.

			Condition 1		Condition 2
Task	From	To	NRIAS 1^st	RIAS 2^nd	RIAS 1^st
4-A	Track #11	Corner	159s	87s	88s

Significant performance differences were found both for the NRIAS 1^st condition and the average overall performance (p<.0003 and p<.0003 respectively). In this walk, no specific route was mentioned; they were just told to go to the corner that they had previously visited.

Figure 3. 7 Transfer Task 4 Path of Travel

Previously, they had gone out the main entrance to reach this corner. There were doors near the end of the station that would be a shortcut, although they had never used them or been told about them. Three out of 15 (20%) of those using their own methods were able to use this shortcut to the corner. Those that used RIAS appear to have learned about the existence of these doors while performing the previous task as 24 out of 30 (80%) were able to find and use these side doors to make a shortcut to the corner. Finding and using paths never used before is quite an accomplishment for many blind people.

Condition 1 Condition 2

Task From To NRIAS 1^st RIAS 2^nd RIAS 1^st

4-B Corner Corner 24s 16s 15s

This street crossing on 4^th Street was not as difficult as the one on King Street. The cars traveled much slower, and almost all made turns in front of the pedestrian. Except for a one-lane bus route on the far side, it was mostly a one-way street in front of the pedestrian. The differences in performance were highly significant for both the NRIAS 1^st condition (p<.001) and the RIAS versus NRIAS results (p<.00005).

Condition 1 Condition 2

Task From To NRIAS 1^st RIAS 2^nd RIAS 1^st

4-C Corner Pay Phone 110s 58s 65s

This task was made difficult by the fact that the pay phone was inside a glass-enclosed bus shelter. There was no outside tactile evidence as to where it was located. Significant performance differences were found both for the NRIAS 1^st condition and the overall performance of RIAS versus NRIAS (p<.025 and p<.006, respectively). Two subjects out of 15 in the NRIAS condition had to ask for help. Four out of 15 subjects using their regular method “timed-out” and could not find the phone. One person out of 30 using RIAS could not find the phone.

Condition 1 Condition 2

Task From To NRIAS 1^st RIAS 2^nd RIAS 1^st

4-D Pay Phone Bus Shelter 71s 0s 0s

The RIAS transmitter at this location identified both the phone and the fact that it was a stop for the #15 bus line. Those without the system had to continue their search to find the correct bus shelter. Seven of the 15 subjects using their normal skills had to ask others to get a positive identification of the proper bus shelter. In addition, two subjects without the system found the correct shelter but reported they “were not sure” if it was for the correct bus line. All of the RIAS users knew they were already at he bus shelter, and, therefore, no extra search time was needed.

Condition 1 Condition 2

NRIAS 1^st RIAS 2^nd RIAS 1^st

Tasks #4 Total 364s 161s 168s

For the entire trip from Caltrain track 11 to the bus shelter for line #15, the results were highly significant for both the NRIAS 1^st – RIAS 2^nd condition and the overall results (p<.00001 and p<.000002, respectively). For the 15 subjects who attempted the four sub-tasks with their regular method, there were a total of 10 tasks that they could not finish and were “timed out.” Of the 30 subjects attempting the same four sub-tasks, only two were “timed out.”

The order of the RIAS condition was shown not to be important in this task. People performed just as well if they used the system first or second. There was no significant difference based on order of use. The t-test showed (p< .43).

3.1.5.1. Time Penalties and Accessibility

3.1.5.1.1. Bus Stop for Route #15

The difficulty of finding bus stops has been confirmed in difficulty ratings in this and previous studies. There is no consistent pattern of their placement, and some are merely placed on pre-existing poles, such as streetlights. Field tests have also showed how difficult these tasks are (Marston & Golledge, 1998b) . Crandall et al. (1996) and Bentzen et al. (1999) found that not one of their subjects could find a bus stop pole even when it had tactile information about the stop. The bus shelter in this experiment had no tactile information about which bus stopped there, and there was another shelter nearby, further confusing the test subjects. The start point for this task was adjacent to the shelter, but with no way to identify which bus stopped there,

they had to search for help or other information. To avoid calculating a penalty with a FSU user time of zero, four seconds was used to indicate the time it took to move to the front of the bus shelter. It took 1685% longer to find and identify the proper bus stop than the baseline FSU. With RIAS, subjects knew exactly which bus stopped there, and all subjects completed the task as fast as the FSU. The t-test statistics for the two conditions showed a significant difference (P<.0009). For those subjects with no vision, t-tests showed a significant difference (P<.006).

Figure 3. 8 Excess Time using Regular Methods and RIAS - Task 4

3.1.5.1.2. Outside Phone

The phone was located about half way down the street from the start point, and those subjects who followed the curb could run into it. There was a wastebasket in front of the phone, and this obstacle slowed many people. The subjects that used their normal aids took 289% longer than the FSU, and those who used RIAS took 137% longer. Unlike the phone in the terminal, subjects had not been to this location previously and there was never anyone using it to give auditory cues. Therefore, this amenity was categorized as one with few cues.

The t-test statistics for the two conditions showed a significant difference (P<.029). For those subjects with no vision, t-tests showed a significant difference (P<.003).

3.1.5.1.3. Walk to Corner #2

Subjects were asked to walk to the first corner that they had visited in the experiment. Although finding information about the street corner might be quite difficult, the task of finding a corner is something in which most blind people are well trained. Subjects made previous trips to the corner by going out the main exit of the terminal. There were some side doors that allowed a short cut to the location. Because the subjects had never used that route, their time was compared to the FSU taking the longer but familiar route. For the subjects who used their regular navigation aids, it took them 78% longer than the FSU. Users of RIAS were usually able to detect the doors for the shortcut, and their average time was equal to that of the FSU taking the longer route. If subjects’ times are compared to the FSU taking the shortcut, the regular aids subjects took 429% longer and the RIAS subjects took 200% longer (see Section 4.6.1 , Spatial Knowledge Revealed by Navigation and Wayfinding Tasks for more information about making shortcuts).

The t-test statistics for the two conditions showed a significant difference (P<.001). For those subjects with no vision, t-tests showed a significant difference (P<.02).

3.1.5.1.4. Medium Difficulty City Street #1

Crossing 4^th Street was much easier than King Street. On the north side was a one-lane dedicated bus lane, which was usually vacant. The south had two lanes with a third turn lane at the corner. This was a much more typical congested city street, with cars parked in front of the terminal. The block was quite short and also had a traffic signal at the other end, so there was no high-speed traffic. Because of the bus lane, there were no vehicles turning onto the street, and few cars went straight across King. Most cars turned right at the terminal. It took the subjects with their regular navigation 58% longer than the FSU. Subjects using RIAS took, on average, no longer than the FSU.

The t-test statistics for the two conditions showed a significant difference (P<.0006). For those subjects with no vision, t-tests showed a significant difference (P<.001).

Figure 3. 9 Transfer Task 5 Path of Travel

3.1.6. Transfer Task 5: Bus Shelter #15 to Track 3

			Condition 1		Condition 2
Task	From	To	NRIAS 1^st	RIAS 2^nd	RIAS 1^st
5-A	Corner	Corner	23s	16s	15s

The street crossing on 4^th Street was not as difficult as the one on King Street. The cars traveled much slower, and all traffic, except for buses, was in one direction. Therefore, it was much easier to hear when the cars stopped. The differences in performance were highly significant for both the NRIAS 1^st condition and the overall performance of RIAS versus NRIAS (p<.001 and p<.0001, respectively). The RIAS gave immediate confirmation that it was safe to cross the street and also gave a directional beam to follow in order to stay in the crosswalk. One subject without the RIAS made an unsafe attempt to cross the street.

Corner – Ticket Window = Guided Walk

			Condition 1		Condition 2
Task	From	To	NRIAS 1^st	RIAS 2^nd	RIAS 1^st
5-B	Ticket Window	Hot Dog Stand	73s	26s	34s

The differences in performance were highly significant for both the NRIAS 1^st condition and the overall performance (p<.004 and p<.0004, respectively). Seven subjects without the RIAS asked for outside help a total of 10 times.

			Condition 1		Condition 2
Task	From	To	NRIAS 1^st	RIAS 2^nd	RIAS 1^st
5-C	Hot Dog	Track #3	126s	63s	60s

Because there were no accessible signs on the track doors, it was difficult to find the correct track. Six subjects using their normal skills had to ask for outside help a total of nine times. Again, both the NRIAS 1^st condition and the overall performance using RIAS were significant (p<.004 and p<.0002, respectively).

	Condition 1		Condition 2
	NRIAS 1^st	RIAS 2^nd	RIAS 1^st
Task #5 Total	222s	105s	109s

For the entire trip from the #15 bus shelter to Caltrain track 3, the results were highly significant for both the NRIAS 1^st - RIAS 2^nd condition and the overall performance (p<.0003 and p<.00002, respectively). For the 15 subjects who attempted the five sub-tasks with their regular method, there were a total of five tasks that they could not finish and were “timed out.” The 30 subjects attempting the same sub-tasks with RIAS had no tasks that were “timed out.”

The order of the RIAS condition was again shown not to be important in this task. People performed just as well if they used the system first or second. There was no significant difference based on order of use. The t-test showed (p<.43).

3.1.6.1. Time Penalties and Accessibility

Figure 3. 10 Excess Time using Regular Methods and RIAS - Task 5

3.1.6.1.1. Track Door #3

In their third and final attempt to find one of the track doors, subjects walked from the hot dog stand to track #3. Finding this unlabeled door took 498% longer than the baseline time. For those who used RIAS, it took them 186% longer. There was a significant difference between the first-time attempts (P<.003). For those with no vision, the results were also significant (P<.005).

3.1.6.1.2. Hot Dog Concession

It was a short walk from the ticket window to the hot dog area, but it was placed so close to the front exit that it seemed to confuse the subjects. It was about 15’ from the other two concessions that they had visited. It took the subjects using their normal navigation aids 462% longer than the FSU baseline, and the subjects who used RIAS took only 160% longer. At times there were voices at the counter to give some cues, so this amenity was categorized as one with few cues.

The t-test statistics for the two conditions showed a significant difference (P<.01). For those subjects with no vision, t-tests showed a significant difference (P<.01).

3.1.6.1.3. Medium Difficulty City Street #2

Walking south across 4^th Street was a bit harder than going the other direction. The turn lanes were at the opposite side of the street; so auditory cues from the street were a bit harder to pick up than when the right turns and traffic were directly in front of the subjects. Those subjects who used their regular navigation skills took 67% longer, and the RIAS users took only 6% longer than the FSU baseline.

The t-test statistics for the two conditions showed a significant difference (P<.002). For those subjects with no vision, t-tests showed a significant difference (P<.0004).

3.1.7. Totals for all Five Transfer Tasks

The degree of efficacy when using RIAS to enable blind and vision-impaired travelers to navigate in large and confusing urban transit environments has been shown here to be highly significant, and system use adds to safety, speed, and spatial knowledge. The results of the five transfer tasks show that this type of system is very beneficial to blind travelers. The total times for the five tasks are shown below. This might represent a normal day for a person making five transfers to different modes. The ability to travel with increased efficiency in a timely and direct manner, complete more tasks, not having to locate people and ask for help, and being able to easily and safely cross busy streets gives people with vision impairments a much better chance to access and use the urban environment. It allows them to achieve more equal access to transit and public buildings in a safe, dignified, and independent manner.

3.1.7.1. Total Travel Task Time

	Condition 1		Condition 2
	NRIAS 1^st	RIAS 2^nd	RIAS 1^st
All 5 Tasks Total	2189s	1129s	1261s

For the 15 subjects who completed all five transfer tasks using NRIAS 1^st and RIAS 2^nd, the results were highly significant (p<.0000002). The average time for NRIAS 1^st was about 36 minutes, and the time fell to 19 minutes with RIAS. All 15 subjects who tried their regular methods first ( NRIAS ) saved time when they tried RIAS 2nd, but RIAS helped the slower test subjects the most. One subject saved 28 minutes and two saved 27 minutes. For the fastest subjects, who all had the ability to see objects, and did not use a mobility aid,RIAS helped reduce their time by one, six, and eight minutes. Data points were plotted for each subject, with their NRIAS 1^st value and their RIAS 2^nd value. A regression line of best fit showed a high correlation effect of R²=.73. When one subject was removed from the analysis because of very inferior navigation skills, the results for the remaining 14 subjects showed a value of R² = .85, indicating a fairly constant effect. They all saved time with RIAS but the high R² value showed that users were consistently slow or fast, relative to the condition mean, whether they used RIAS or NRIAS.

The benefits of RIAS appears so powerful that there is no significant difference between those that used the system for their first trail and those that had first tried the tasks on their own and then tried the experiment again with the RIAS. The t-test value showed that the order was not significant (p<.25).

People found locations quicker and missed them less often when using RIAS them when using their regular methods. This was achieved with only 10 to 15 minutes of training. Street crossing results showed that, without the system, many people made potentially fatal decisions and that there was much hesitation and even some refusals to cross dangerous streets. In all, subjects using their own skills made 38 attempts to cross the street when it was unsafe to do so.

These five tasks were designed to approximate a typical day’s transfer tasks for a daily urban traveler. The travel times for the RIAS 1^st condition was fully 39% less than for those using NRIAS first. When those using NRIAS first tried the RIAS, their times fell, on average, by 49%. This is a tremendous saving in effort and personal stress. The times would certainly drop even more with repetition and learning. But even in a novel environment, the ability to save 49% of the normal time of these tasks, and the increased completion rate, are a great incentive for more and safer travel.

A sighted research assistant (FTSU) who had never been to the site received the same instructions, and it took him 9.47 minutes to complete the route on his first attempt (see Section 1.6.6 , Sighted Subjects for Baseline ). The 15 vision-impaired subjects who tried their regular method first took, on average, 36.48 minutes. The time “penalty” for vision loss was thus 3.85 times more effort than for the sighted. This penalty shows that to date there is no “equal access” to transit. The average time for those who used RIAS first was 22.13 minutes. Their penalty fell to a more tolerable and equitable 2.34 times the time for the sighted.

There was a wide range of subjects with various skills and degrees of vision loss. If only the top performers on these transfer tasks are compared, some very revealing evidence for RIAS is uncovered. Of those 15 that used RIAS first, six (40%) had times that were less than twice as long as the FTSU baseline data. The best time was only 9% longer than the baseline, with the next five having times of 21%, 24%, 67%, 85%, and 88% longer. That is certainly more like the reasonable accommodations and equity that is the focus of the ADA. When the results of those vision-impaired people who used RIAS for their second trial are compared against a fully sighted first-time person, the results are even more powerful. Nine of 15 subjects (60%) had times within twice that of the sighted baseline. One person actually completed the task 1% faster than the sighted subject. The next lowest eight times were 7%, 30%, 31%, 33%, 41%, 51%, 57%, and 68% longer.

The possible savings of so much time, effort, and stress shows that the ability to identify locations and access directional cues is quite helpful in providing increased access to transit and public buildings for the vision-impaired.

3.1.8. Unsafe Attempts to Cross Street

Street crossing can be very unsafe for a blind pedestrian. With proper training, this group can perform amazing and fearless (to the sighted) feats of mobility. However, many intersections are not easy to cross. Irregular angles of intersection, and configuration and timing of turn lanes and traffic flow can make many intersections quite difficult. In the NRIAS condition, subjects crossed streets 80 times and made 38 unsafe (48%) attempts to cross those streets while the WAIT light was on and traffic had the right-of-way. At the most dangerous crossing, Task 2-A, three subjects waited their full four minutes and did not cross the two-lane street. If there had been no researcher watching for traffic or helping them across, this one intersection bottleneck could have completely halted any further travel progress or resulted in bodily harm. At this same intersection, fully 13 of 25 subjects (52%) attempted unsafe crossings with their regular method a total of 20 times. In addition, 17 out of 80 (21%) attempts to cross missed the opposite curb, also putting them in danger. When using RIAS, not one unsafe attempt was made to cross the street, because the receiver told them the status of the WALK and WAIT signal. Only one person using RIAS missed the opposite corner. Figure 3.11 shows the wide and narrow beams at the intersection that give the traveler both street information and specific information about the phase of the WALK signal and other interaction information. Independence, environmental information, and trip making enhancements are wonderful outcomes from using RIAS, but the safety of the blind pedestrian is a major benefit of this system.

Figure 3. 11 Oblique View of RIAS Installation at King and 4th Streets

Often, the only way a vision-impaired person can navigate about an environment is to search for a sighted person and then ask for help. This makes some people feel vulnerable and dependent on others. Many objects in the environment are not marked in any fashion, and, without vision, there is no way to differentiate objects such as a bank of doors. Often there are few, if any, people around to ask, and some of them might not know the answer, refuse to help, or be unable to speak the same language. Fear of personal assault makes some people want to avoid drawing any undue attention to their vulnerability as a blind person. These fears are yet another factor that might keep people from making the trips they desire and might negatively impact their ability to enjoy full and active travel and activity participation. Many subjects in this experiment, when using their regular method of travel, had to ask people for help, and often there was no feedback to help identify locations.

There were 75 sub-tasks for the NRIAS 1^st condition in Task 1 and these subjects asked for assistance from others 23 times (31%) of the time.
There were 75 sub-tasks for the NRIAS 1^st condition in Task 2 and these subjects asked for assistance from others 35 times (47%) of the time.
There were 45 sub-tasks for the NRIAS 1^st condition in Task 3 and these subjects asked for assistance from others 22 times (49%) of the time.
There were 60 sub-tasks for the NRIAS 1^st condition in Task 4 and these subjects asked for assistance from others 9 times (15%) of the time.
There were 45 sub-tasks for the NRIAS 1^st condition in Task 5 and these subjects asked for assistance from others 18 times (40%) of the time.

Since subjects knew this was a test and that the researcher was with them, they must have felt safer than if they were truly on their own. In a real situation, some of these people would have probably not bothered to risk their safety and ask for help, giving up on the task instead. When subjects used the RIAS, not one person asked for help. In fact, two people were offered help by strangers and they politely refused, not needing any assistance. What is even more revealing is the fact that, for those who tried RIAS first and then used their regular method, there were only three requests for assistance out of 100 sub-tasks (3%). Having found the locations first using RIAS, many questions of identity or location had already been answered. These data indicate how dependent a blind traveler is on other people and how vulnerable thez might be in an urban environment. Having to rely on others for simple verification of objects and directions can be a heavy penalty to pay. This reliance on others contradicts attempts to promote access or independence.

BACK TO OVERVIEW

BACK TO TABLE OF CONTENTS

NEXT SECTION