jjj

Detecting Approaching Vehiclesat Streets with No Traffic ControlRobert Wall Emerson and Dona Sauerburger

Abstract: This study assessed the ability of people with visual impairments toreliably detect oncoming traffic at crossing situations with no traffic control. Inat least one condition, the participants could not hear vehicles to afford a safecrossing time when sound levels were as quiet as possible. Significant predictorsof detection accounted for a third of the variation in the detection time.

For pedestrians with visual impairments(that is, those who are blind or have lowvision), crossing a street where there is notraffic control can present a challenge.Crossing situations with no traffic con-trols include residential intersectionswhere only one street has a stop sign,channelized right-turn lanes with no sig-nal, midblock locations, and roundabouts.In these situations, pedestrians must crosseither when drivers yield or in a gap intraffic. Because drivers often do not reli-ably yield to pedestrians, even those withwhite canes (Geruschat & Hassan, 2005;Guth, Ashmead, Long, Wall, & Pon-chillia, 2005; Inman, Davis, & Sauer-burger, 2005; Sauerburger, 2003), pedes-trians often must cross in a gap in traffic.For a pedestrian to cross in a gap in traf-fic, gaps must exist that are long enoughto allow time to cross, and the pedestrian

must be able to recognize when thesegaps exist (Sauerburger, 1999).

Pedestrians who are visually impaireduse their hearing to detect approachingvehicles and gaps in traffic. Strategies forusing hearing to cross streets with no traf-fic control were first developed in the late1940s (Wiener & Siffermann, 1997).Early orientation and mobility (O&M) in-structors used demonstrations to convincetheir students, veterans who were blind,that streets were clear to cross wheneverthe environment was quiet (Sauerburger,1999). By measuring the time from whenthe veteran heard a vehicle approachingto when the vehicle arrived, they deter-mined that all vehicles could be heard farenough away to know that it was clear tocross whenever it was quiet. For pedes-trians who are blind, the strategy of “crosswhen quiet,” for streets where there is notraffic control, continues to be used today(Allen, Courtney-Barbier, Griffith, Kern,& Shaw, 1997; Jacobson, 1993; LaGrow& Weessies, 1994; Pogrund et al., 1993).

In the traffic environment of the1940s and 1950s, the strategy of “cross-ing when quiet” was probably effective,but decades later, Sauerburger (1989,

This project was supported by GrantR01EY12894 from the National Eye Institute.The content of this article is solely the respon-sibility of the authors and does not necessarilyrepresent the official views of the NationalEye Institute or the National Institutes ofHealth.

©2008 AFB, All Rights Reserved Journal of Visual Impairment & Blindness, December 2008 747

1995, 1999, 2006) and Snook-Hill andSauerburger (1996) observed that therewere situations in which, even when itwas quiet, approaching vehicles could notbe heard well enough to be detected withsufficient warning to know whether therewas a gap in traffic that was long enoughin which to cross. In the 1940s, the strat-egy of “crossing when quiet” was effec-tive because vehicles were louder. Priorto 1972, the sound level of vehicles in theUnited States was commonly 90 decibels(dB) (Wiener & Lawson, 1997), but theU.S. Federal-Aid Highway Act of 1970and the Noise Control Act of 1972 di-rected the Federal Highway Administra-tion and the Environmental ProtectionAgency, respectively, to develop standardsfor mitigating highway traffic noise, andcurrently automobile manufacturers in theUnited States design cars to meet the Eu-ropean noise limit of 78 dBA (FederalHighway Administration, 1995). Becausethe dB scale is logarithmic, observers per-ceive a decrease of 10 dB as a halving ofthe sound level, so the sound of passengercars was reduced to less than half of whatit was before 1972. The trend toward qui-eter vehicles is expected to continue ashybrid and electric vehicles become moreprevalent (Wiener, Naghshineh, Salis-bury, & Rozema, 2006). The decrease inthe general sound level of vehicles de-mands that the strategy of crossing whenquiet be investigated again to establish itsusefulness.

Sauerburger (2006) discussed strate-gies for teaching pedestrians who areblind how to recognize when it is notpossible to hear or see the traffic suffi-ciently far enough away to know when itis clear to cross at streets with no trafficcontrol. Scheffers and Myers (1995) also

attempted to define parameters wherebypeople who are blind would be recom-mended not to cross streets independentlywith no traffic control. Their guidelinesare that when there is “all quiet” (that is, nomotor or “masking sounds” are present) or“all clear” (that is, motor or masking soundsare only in the distance), blind pedestrianswith a good pace and judgment are able tocross without assistance only if all the fol-lowing conditions are met: The traffic is“light” (0–10 vehicles per minute), thestreet is “narrow” (no more than one drivinglane and one parking lane in each direction),the traffic speeds are “slow” (approximately25 miles per hour, mph, or less), and thereis good visibility (half a block or more,considering illumination, hills, windingroads, and obstruction of view from largevehicles). In addition to parameters that in-fluence the safety of crossing when quiet,masking sounds will influence a pedestri-an’s ability to hear approaching traffic. Forexample, traffic noise at busy roundaboutscan compromise the ability of a pedestrianto hear approaching traffic well enough toknow that it is clear to cross a street(Wadhwa, 2003).

In view of concerns about the abilityof pedestrians with visual impairmentsto rely on the strategy of crossing whenquiet and about the effect of maskingsounds, it would be helpful to study thefactors that may affect the ability ofindividuals to hear approaching vehi-cles well enough to know there is a gaplong enough in which to cross a street.The study presented here considered towhat degree, if any, the auditory detec-tion of approaching vehicles was af-fected by the level of ambient sound;the sound level and speed of approach-ing vehicles; and physical features of

748 Journal of Visual Impairment & Blindness, December 2008 ©2008 AFB, All Rights Reserved

the environment, such as hills, bends inthe road, trees, and obstacles. One goalof the study was to determine whetherthere are conditions in which it is notpossible for a person with good hearingand an average walking speed to hearapproaching vehicles with enoughwarning to know that it is clear to crossa street when it is quiet.

MethodsPARTICIPANTS

The participants (17 women and 6 menwith visual impairments) ranged in agefrom 24 to 67 (M � 46.8). Nine hadcongenital vision loss, and 13 lost theirvision at age 3 or older. The etiologies oftheir visual impairments included retinop-athy of prematurity, glaucoma, retinal de-tachment, amaurosis, optic chiasm tu-mors, sarcoid growth, injury, retinitispigmentosa, diabetic retinopathy, retinalblastoma, macular degeneration, and un-known. Five participants had more thanlight perception, but all wore blindfoldsduring the data collection.

The participants were recruited by thesecond author through contacts in Mary-land and Virginia. Potential participantswere informed that normal hearing was arequirement of the study, and they all hadtheir hearing screened on their arrival atthe testing site. A portable audiometerwas used to assess the potential partici-pants’ hearing at 1,000, 2,000, 4,000, and8,000 Hz. As long as a participant did nothave a hearing loss greater than 20 dB atany of these frequency bands, it was deter-mined that he or she had hearing in thenormal range. One participant was elimi-nated because of hearing loss. All the par-ticipants were independent travelers who

crossed streets often as part of their dailytravel.

PROCEDURES

We identified potential data collectionsites in residential areas within the SilverSpring, Maryland, area. Each site had tobe a street that a pedestrian would logi-cally be expected to cross. For example,one site had a bus stop on one side of thestreet and an apartment complex on theother. Each site also had to have vehiclesapproaching at least 100 yards from thenearest traffic control device (such as atraffic light, stop sign, or yield sign) onthe street being used to assess the detec-tion of gaps in traffic. In addition, eachsite had to have an ambient noise levelthat was equal to or less than the averagelevel of “quiet” for residential intersec-tions in that area.

To determine this average level ofquiet, sound-level readings of minimal re-sidual ambient sound (levels of relativequiet) were taken at 12 sites in the resi-dential study area in the suburbs of SilverSpring. Comparison sound levels werealso taken in downtown Atlanta and res-idential and business areas in Annapolis,Maryland (areas that were representativeof a large and a small city). Sound levelswere taken in moderately busy traffic ar-eas in the middle of the day. The averageminimal residual sound level for “resi-dential or urban collector” roadways inthe area in which the study was conductedwas 44 dBA (range 37 to 49.5), the aver-age sound level was 49 dBA (range 45.5to 53) in Annapolis, and the averagesound level was 66 dBA (63.5 to 69) inAtlanta. Note that all sound-level mea-surements that were used in this studywere A weighted (dBA), which means


that the sound-pressure scale was ad-justed to conform to the frequency re-sponse of the human ear. Thus, the mea-sures more closely reflect the loudnessthat a human listener would experience.

Three sites were chosen that met thequalifications just discussed, reflected dif-ferent environmental conditions, and con-formed to the average residual soundlevel of less than 44 dBA. All three siteswere on two-way roadways with one lanein each direction near a place where res-idential streets with no traffic control in-tersected with streets with stop signs. Site1 was a T-shaped intersection, with eachapproach along the main street to thecrossing point being straight and clear ofobstructions for approximately 1,000 feet.On half the trials at this site, the experi-menters blocked sound on one side of theparticipants with an 8-foot-wide and5-foot-high barrier that simulated aparked vehicle blocking the sound of on-coming vehicles. Site 2 was a 90-degreesquare intersection; approximately 150feet to the right of the crossing point, themain street had a severe bend in the road-way, beyond which the approaching ve-hicles could not be seen, and approxi-mately 100 feet to the left was a minorbend that was more gradual (approachingvehicles could be seen approximately 200feet away). Site 3 was a T-shaped inter-section with a steep hill that droppedaway from the crossing point starting atapproximately 50 feet to the right and aheavily treed area that obscured oncom-ing traffic beyond approximately 100 feetto the left.

Data were collected at crossing pointswhere there was no traffic control on thestreet to be crossed but where a pedestriancould logically choose to make a cross-

ing. The participants were not asked ac-tually to cross the street. Two to threeparticipants at a time sat one behind theother at the side of the road. The partici-pant closest to the roadway was alwaysapproximately 2 feet back from the curb,and the other participants were 1 footbehind the person in front of them. Theparticipants sat throughout the data col-lection to reduce their fatigue and maxi-mize their concentration on the sound ofapproaching vehicles. Their task was toindicate, by raising their left or right hand,when they first heard a vehicle approach-ing from the left or the right. The partic-ipants used minimal hand movements tolimit noise that might influence the otherparticipants’ decisions.

The participants remained at each datacollection site for approximately 30 min-utes of data collection. Breaks were takenwhenever any participant indicated theneed for a respite. Water, snacks, andpillows were provided to maximize theparticipants’ comfort and focus on thetask. The weather during all the data col-lection times was fair (sunny or partlycloudy) with no rain and no significantwind.

Data were collected by videotape. Oneexperimenter remained with the partici-pants to answer questions and ensure thatall remained on task, while the othermanned the equipment. A video camerawas set up so that both the participants’hand movements and the presence ofpassing vehicles could be identified. Alsoin the frame of the camera was the read-out for a Cel-490 sound-level meter (Ca-sella) and a speed gun (Sports Radar).The sound-level meter continuously mea-sured the sound level of the environmentin dBA. The speed gun automatically


activated when a vehicle came into itsrange from either approach (left or right).All the equipment was set up by the sideof the road, facing the participants. Toensure that the presence of the equipmentdid not have a significant effect on thebehavior of drivers, data on speed werecollected at similar times of day, withonly one person and the speed gun placedin a hidden location. No significant dif-ference in speeds was noted.

The general sound level in each listen-ing environment varied with a normalcourse of gusts of wind, traffic in thedistance, people passing on the street, andother environmental factors, but loudmasking noises like leaf blowers andlawnmowers were deemed to be too in-trusive and so were not allowed to be apart of the data collection process. In theanalyses, reference to “quiet” status inany environment refers to the point atwhich the ambient level of sound in thatenvironment was as low as it was likely toget. Across the listening conditions, thislevel of sound generally corresponded to42 dBA.

The issue of what is quiet can be com-plex. Presumably, it is quiet when theambient sound level is low, and there areno extraneous sounds. However, the levelof ambient sound varies during the dayand between communities. In 1971, WyleLaboratories recorded the sound levelsduring 24-hour periods in a variety ofcommunities and observed a fairly con-stant lower level of sound that alwaysremained when sounds such as vehicles,barking dogs, airplanes, trains, lawnmow-ers, and conversation, had died away.This residual noise level varied greatlybetween locations and during differenttimes of the day. In cities, the daytime

residual noise level ranged between 62dBA and 69 dBA, and for urban or sub-urban residential areas, it ranged from 42dBA to 56 dBA.

The ambient sound level in this exper-iment was consistent across days, times ofday, and environments. If the ambientsound level dropped below 45 dBA, itwas deemed to be quiet. This level ofsound was experientially valid, since atthese times all that was heard was perhapsa gentle rustling of leaves or a bird in thedistance. Any noise at all would bring theambient sound level above this point. Asa check on this operational definition of“quiet,” the experimenter who could notsee the sound-level meter indicated withhand signals when she thought the ambi-ent level of sound had reached the quietpoint for that location (that is, the noiselevel had gotten as low as it could go).The experimenter’s signals correspondedwith a sound level reading of 42 dBA to44 dBA for 97% of the trials.

MEASURES

Of primary interest in this study was howwell the participants detected traffic usingtheir hearing. Toward this end, the prin-cipal measure used in the analysis was thetime from the first detection of an ap-proaching vehicle to the time when thevehicle passed in front of the pedestrian.A secondary measure of “safety margin”was defined as the difference between theestimated crossing time and the time fromwhen the vehicle was detected to whenthe vehicle arrived. We estimated thecrossing time using the minimum stan-dard road width for urban collectors. Thewidth for a “residential or urban collec-tor” roadway was used because these aresituations in which a pedestrian may often


be required to cross the street and wherea moderate amount of traffic usually ex-ists. All the sites used in the study satis-fied the criteria for being an urban collec-tor. The minimum standard width forurban collectors was 10 feet for each laneand 4 feet for each shoulder (AmericanAssociation of State Highway and Trans-portation Officials, 2004). Thus, assum-ing a two-way road, the crossing distanceused in this study was 28 feet. The timerequired to cross the roadway was there-fore calculated to be 7 seconds, on thebasis of a crossing speed of 4 feet persecond, which is commonly used as theaverage walking speed of pedestrianscrossing streets (LaPlante & Kaeser,2004). If a safety margin was negative,the vehicle reached the crossing point be-fore the participants would have com-pleted the crossing if they had started tocross just before they heard the vehicle. Ifthe safety margin was positive, the par-ticipants would have finished their cross-ing and stepped out of the roadway beforethe vehicle passed the crossing point. Anywider road widths would have the effectof decreasing any safety margins thatwere found.

Attempts were made to design the datacollection so as to maximize the earliestpossible detection of approaching vehi-cles and to be as conservative as possiblein the ability to disprove the hypothesisthat there are quiet situations in which it isnot possible to hear vehicles far enoughaway to afford a safe crossing. More thanone participant at a time indicated whenthey first heard a vehicle approaching,and the resulting analyses used only thefirst response given within each group ofparticipants. Having more than one per-son respond to each oncoming vehicle

controlled for momentary lapses in atten-tion by one or two participants, but stillallowed for the early detection of all ve-hicles. Having participants seated duringthe data collection reduced their fatigueand increased their concentration on thetask. Finally, the calculation of safetymargins, which involved an assumptionof crossing time, used the minimumcrossing distance for the type of roadwayused in the study. Urban collector road-ways, of the type used in this study, havea range of typical lane widths (AmericanAssociation of State Highway andTransportation Officials, 2004). By us-ing the smallest lane width, we obtainedthe most conservative safety margins.If, after designing the study to allow forthe earliest detection of vehicles and thequickest crossing time, we still foundinstances in which the participantscould not hear vehicles far enough awayto afford a safe crossing, the resultswould be that much more meaningful.

ResultsThere were 6 environmental conditionsunder which the participants made detec-tion decisions of approaching vehicles: astraight roadway with no obstacles(“straight”), a straight roadway with abaffle to obstruct the sound (“baffle”), asevere bend in the road, a minor bend inthe road, a hill falling away from theparticipants, and a heavily treed ap-proach. Only the first of a group of par-ticipants to respond to an oncoming ve-hicle was used as the detection time forthat vehicle. Taking only these “first re-sponders” into consideration, there were805 detection trials across the 6 environ-mental conditions: 119 with the baffle,290 on the straight roadway, 123 on the


severe bend, 85 on the minor bend, 124 onthe hill, and 64 from the treed approach.

Across all 805 trials, the detection time(the time from when the vehicle was de-tected to when the vehicle passed in frontof the participants) ranged from 2 secondsto 35 seconds (M � 9.63, SD � 4.94,median � 9.00). Across all environmen-tal conditions, the average safety marginwas 2.63 seconds (SD � 4.94), and themedian was 2.0 seconds. Safety marginsranged from -5 seconds to 28 seconds. Asafety margin of -5 seconds meant that avehicle would have arrived at the crossingpoint 5 seconds before the crossing wascompleted if the crossing had been startedjust before the person heard the vehicle.

Across all 805 trials, the ambient soundlevel just before the next oncoming vehi-cle was detected ranged from 36 dBA to60 dBA (M � 44.7, median � 44). Of the805 trials, 360 had ambient sound levelsthat were considered quiet when the par-ticipants were listening for the next ap-proaching vehicle. These trials representthe best possible listening condition. Ta-ble 1 shows the range and average detec-tion times of approaching vehicles duringtimes of quiet. The straight, unimpededcondition had no detections less than 7seconds, whereas the severe bend and thehill had the lowest average detection

times. Because of unequal variancesacross groups, it was not possible to con-duct an overall analysis of variance. How-ever, a Scheffe test comparing the detec-tion times for each pair of conditionsshowed significant differences betweenthe straight condition and both the severebend and hill conditions, between the baf-fle condition and both the severe bend andhill conditions, and between the minorand severe bends.

As Table 1 shows, half the safety mar-gins in the severe bend condition were 0or less (the median safety margin was 0).This finding means that, in this condition,if the participants had started crossing justbefore they heard a vehicle, half the timethey would not have been able to com-plete their crossing before the vehicle ar-rived at the crossing point. Note that thisis true on the assumption that listeningand crossing are performed under ideallistening conditions. It may be argued thata safety margin of 0 means that a personis relatively safe because he or she isfinishing the crossing just as the next ve-hicle passes the crossing point. However,most pedestrians have a second or two ofadded crossing time to make the decisionto cross and to begin the motion to cross.We argue that a safety margin of 0 oftenreflects the situation in which a pedestrian

Table 1Detection times when listening in quiet.

Listeningconditions

Totaltrials

Range ofdetection times

(in seconds)

Average (SDs inparentheses) detection

time (in seconds)

Average (SDs inparentheses) safetymargin (in seconds)

Median safetymargin

(in seconds)

Straight 144 7–35 13.79 (4.67) 6.79 (4.67) 6.00Baffle 50 5–32 13.34 (5.73) 6.34 (5.73) 5.00Severe bend 60 4–19 7.62 (3.11) 0.62 (3.11) 0.00Minor bend 39 4–28 11.08 (4.93) 4.08 (4.93) 3.00Hill 43 4–20 9.51 (3.56) 2.51 (3.56) 3.00Treed approach 24 7–26 11.25 (4.40) 4.25 (4.40) 3.50


is still somewhat in the final lane of trafficwhen the next vehicle passes.

We considered that this effect may bethe result of participants being influencedby the lingering effects of recently ceasedmasking sounds. Masking sounds, such asreceding traffic, can elevate the thresholdof detection for sounds that occur soonafter the masking sound ends and it be-comes quiet, a phenomenon known asforward masking (Elliott, 1971). After thecessation of a masking sound, the lengthof time needed to return to a normalthreshold for detecting sounds is gener-ally about 100 milliseconds. Mediansafety margins in trials for which a par-ticipant had at least 1 second of relativequiet before he or she detected an ap-proaching vehicle showed the same pat-tern of results as in Table 1.

Having examined the possible effect offorward masking, we considered the pos-sibility that long periods of quiet mayhave affected the participants’ detectionabilities because of the participants lossof focus on the task. We used only thestraight condition to help isolate the effectof a longer period of quiet on detectionability. Looking at only those trials in thestraight condition in which the ambientsound level had dropped to 44 dBA orless before the next vehicle was detected(n � 141), we found no correlation be-tween how long the ambient environmenthad been quiet and the next detection time(R2 � .0006).

The differences between environmentalconditions are exemplified by looking ataverage safety margins for the ambientsound level in each environmental condi-tion (see Figure 1). Two panels are usedin the figure to reduce visual clutter forthe reader. In the lower panel, the baffle,

minor bend, and treed conditions showsimilar patterns of decreasing safety mar-gins as the ambient sound level increases.At about 50 dBA, the average safety mar-gin becomes low enough to cause concernabout detectability. The upper panelshows that the average safety margin inthe straight condition consistently re-mains positive up to about 52 dBA. How-ever, detection at the severe bend and hillconditions degrades at much lower soundlevels. The severe bend showed an aver-age negative safety margin at 38 dBA,while the hill began showing negativeaverage safety margins as low as 43 dBA.

Although environmental factors andhow quiet the environment was before theapproaching vehicle was detected seem toaffect how well the participants were ableto detect oncoming vehicles, how loud avehicle was and how fast it was travelingmay also play a role. We conducted astepwise regression analysis on all 805trials to determine which variables hadthe largest impact on the participants’ability to hear oncoming vehicles. Predic-tive factors that were entered into theregression analysis included environmen-tal condition, vehicle speed, vehiclesound level, and ambient sound level be-fore the vehicle was detected by thesound-level meter. Detection time was theoutcome measure. The analysis indicatedthat the ambient sound level was thestrongest predictor of detection time, fol-lowed by the environmental condition,the sound level of the passing vehicle, andthe speed of the vehicle. These factorsaccounted for a third of the variability inthe detection-time measure (R2 � .34).

One may argue that the speed andsound level of the vehicle would be ex-pected to be highly correlated with each


-6

-4

-2

0

2

4

6

8

10

12

35.00 40.00 45.00 50.00 55.00 60.00

Ambient sound level (dBA)

Safe

ty m

argi

n (s

ec)

straight

severe bend

hill

-6

-4

-2

0

2

4

6

8

10

12

35.00 40.00 45.00 50.00 55.00 60.00


Safe

ty m

argi

n (s

ec)

baffle

minor bend

trees

Figure 1. Average safety margin, by ambient sound level in each condition.


other and therefore should not both beentered into the prediction equation.However, these two variables had a non-significant correlation coefficient of .41.Each contributed significantly to the abil-ity of the participants to detect vehicles,but neither was as predictive as the am-bient sound level or the environmentalcondition. These results included vehicleswith sound levels ranging from 56 dBA to89 dBA and speeds from 13 mph to 60mph. The analysis shows that, as onewould expect, the lower the ambientsound level, the better a person can hearapproaching vehicles. Also, to some de-gree, vehicles that are louder or movingmore slowly are detectable earlier. Anal-yses performed within each environmen-tal condition showed the same sequenceof predictor variables.

We analyzed the effect of ambientsound level, vehicle sound level, and ve-

hicle speed on detection time moreclosely. To do so, we used only the datafrom the straight roadway (without thebaffle). This environmental conditionhad no obstructions to hearing in eitherdirection and so allowed for a clearanalysis of the relative effects of ambi-ent sound level, vehicle sound level,and vehicle speed on the detection ofoncoming vehicles. These 288 trialsshowed a strong negative relationshipbetween ambient sound level and detec-tion time (see Figure 2).

In trials on the straight roadway, asthe ambient sound level increased, thedetection times decreased. Once the am-bient sound level is above approxi-mately 50 dBA, it becomes virtually im-possible to hear vehicles far enough awayto know whether it is clear enough to beable to complete a crossing before thevehicles arrive.

R2 = 0.2434

0

5

10

15

20

25

30

35

40

30 35 40 45 50 55 60


Det

ectio

n tim

e (s

ec)

Figure 2. Correlation between ambient sound level and detection time in the straight condition.


In relation to the other factors, thesound level and speed of the approachingvehicles added little predictive value tothe regression equation (R2 � .000003 forsound level and .0316 for vehicle speedwhen correlated with detection time).This finding implies that in a good listen-ing environment (that is, with no obstruc-tions and a quiet environment), a personis able to hear most vehicles well, nomatter what their specific characteristics.The slight upturn in the trend line at thelower vehicle speeds (see Figure 3) indi-cates that the slowest moving vehicles(less than 21 mph) may provide sufficientinformation for detectability. It may do sobecause no matter how noisy or quiet avehicle is (given the current levels of ve-hicle sound output), if it is moving slowlyenough, then pedestrians can hear it withenough warning to know there is time toget across the road if they start when it isquiet. However, it is important to notethat the quietest vehicle on the straightroadway was 57.9 dBA, and only two

vehicles in 288 trials were quieter than 60dBA. Further study is needed to deter-mine how the new generation of hybridand quiet internal combustion vehicleswill affect these results.

DiscussionThe situations in which a pedestrian witha visual impairment is expected to crosswithout the advantage of a walk signal areincreasing, especially in light of the in-creased construction of roundabout inter-sections. The traditional approach tocrossing where there is no traffic controlhas been to cross when it is quiet. Thefindings of this study seem to support theimportance of having as quiet an environ-ment as possible. Even in ideal environ-mental conditions, the level of ambientnoise does not have to increase by muchto affect a pedestrian’s ability to detectoncoming traffic. But this is too simplistican approach to this kind of crossing situ-ation. It is important for pedestrians whoare blind and their O&M instructors to be

0

5

10

15

20

25

30

35

40

0 10 20 30 40 50

Vehicle speed (mph)

Det

ectio

n tim

e (s

ec)

60

Figure 3. Correlation of vehicle speed with detection time in the straight condition.


aware that a pedestrian’s ability to hearoncoming vehicles can vary significantlyand, in some situations, can render thepedestrian unable to determine whetherthere is a gap long enough to cross theroadway, even when quiet.

As in Scheffers and Myers’s (1995)guidelines, the findings of this study sup-port the inclusion of many different pa-rameters in addition to ambient soundlevel. The findings indicate that environ-mental features, such as bends and hills,can significantly affect a pedestrian’sability to hear vehicles well enough tomake safe crossing judgments. However,environmental features and the other fac-tors considered in this study accountedfor only a third of the variation in detec-tion times. This finding implies that ob-serving the parameters considered in thisstudy, such as the presence of hills andbends, ambient sound, and the speed andsound level of the vehicles, will not fullypredict whether a given crossing situationwill constitute a problem. It is imperativethat O&M instructors teach pedestrianswith visual impairments to analyze eachsituation on its own merits and not relyonly on the presence or absence of thefeatures considered in this study to eval-uate whether a given street is crossable.

The results also support the develop-ment of a better understanding of whatconstitutes “quiet” and how ambientsound affects crossing decisions. In all theconditions but one, as the ambient soundlevel started to rise, the participants’ abil-ity to hear vehicles did not begin to de-teriorate until the ambient sound levelrose above 43 dBA. At the hill, however,starting with an ambient sound level aslow as 38 dBA, a rise in the ambientsound level was correlated with a severe

decline in the participants’ ability to hearvehicles. That is, although some condi-tions led to better detection than others, inmost conditions the level of detection didnot substantially decline between ambientsound levels of 38 dBA and 43 dBA. Atthe hill, the ability to hear vehicles wasmuch different at 43 dBA than it was at38 dBA. Thus, in some conditions, thepedestrians’ ability to detect approachingvehicles can tolerate slight increases inbackground sounds, but in other condi-tions, such as the hill, it is severely im-peded by the slightest background noise.Pedestrians who are blind need to recog-nize situations in which a slight noise cansignificantly decrease their ability to hearapproaching vehicles.

Recall that in this study we assumed aminimal standard crossing width for twotraffic lanes with shoulders. Any widercrossing would not affect detection times,but it would affect safety margins. Thepatterns of safety margins shown in thesedata would only become worse with awider crossing, especially at the hill andthe severe bend. Another factor that mayincrease the effects found in the study isthe increase of hybrid and other quietervehicles on the roadways.

In the same way, a general increase inambient sound level will have a concom-itant decrease in detectability. The sam-pling of sound levels in different urbanenvironments that was used to select thestudy location demonstrates how differentenvironments can influence the need fordifferent street-crossing strategies. A dif-ferent strategy may be necessary when apedestrian who is blind is traveling in thenoisier environment of a larger city. InAnnapolis, the average ambient soundlevel of 49 dBA puts it close to the level


of background noise at which it would bedifficult to hear an approaching vehiclewith enough time to know if it is clear tocross. The roundabout in downtown An-napolis is a good example of a situation inwhich a pedestrian who is blind would beexpected potentially to cross in front ofmoving traffic, without the aid of a walksignal and with buildings blocking boththe sightlines and hearing lines of ap-proaching traffic. The trends in our dataindicate that at an ambient sound level of66 dBA, as in downtown Atlanta, pedes-trians would be unable to hear approach-ing vehicles more than about a secondaway. Thus, to avoid unacceptable risks,they would be forced to consider alterna-tives, such as traveling routes that rely onstandard signalized intersections.

In conclusion, the traditional strategyof crossing when quiet remains effectivein some situations, but not in others. Al-though the ability to hear approachingvehicles well enough to cross a street isaffected by such conditions as ambientsound levels, the presence of certain en-vironmental features like hills and bends,and traffic speeds, it is not possible topredict the ability to hear approachingvehicles by noting these features alone.Pedestrians who are blind need to beaware that before they can assume that itis clear to cross when quiet in a givensituation, they must first observe theirability to hear the vehicles in that situa-tion and determine whether they can hearwell enough to know that it is clear tocross.

ReferencesAllen, W., Courtney-Barbier, A., Griffith, A.,

Kern, T., & Shaw, C. (1997). Orientation

and mobility teaching manual (2nd ed.).New York: CIL Publications.

American Association of State Highway andTransportation Officials. (2004). A policyon geometric design of highways andstreets (5th ed.). Washington, DC: Author.

Elliott, L. L. (1971). Backward and forwardmasking. Audiology, 10(2), 65–76.

Federal Highway Administration. (1995).Highway traffic noise analysis and abate-ment policy and guidance. Washington,DC: Federal Highway Administration Of-fice of Environment and Planning, Noiseand Air Quality Branch. Retrieved June29, 2008, from http://www.fhwa.dot.gov/environment/polguid.pdf

Geruschat, D. R., & Hassan, S. E. (2005).Driver behavior in yielding to sighted andblind pedestrians at roundabouts. Journalof Visual Impairment & Blindness, 99,286–302.

Guth, D., Ashmead, D., Long, R., Wall, R., &Ponchillia, P. (2005). Blind and sightedpedestrians’ judgments of gaps in trafficat roundabouts. Human Factors, 47,314 –331.

Inman, V. W., Davis, G. W., & Sauerburger,D. (2005). Roundabout access for visuallyimpaired pedestrians: Evaluation of ayielding vehicle alerting system fordouble-lane roundabouts. In RoundaboutConference Proceedings: 2005, Vail, CO.Washington, DC: Transportation ResearchBoard.

Jacobson, W. H. (1993). The art and scienceof teaching orientation and mobility to per-sons with visual impairments. New York:American Foundation for the Blind.

LaGrow, S., & Weessies, M. (1994). Orien-tation and mobility: Techniques for inde-pendence. Palmerston North, New Zea-land: Dunmore Press.

LaPlante, J., & Kaeser, T. (2004). The con-tinuing evolution of pedestrian walkingspeed assumptions. ITE Journal, 74(9),43–40.

Pogrund, R., Healy, G., Jones, K., Levack, N.,Martin-Curry, S., Martinez, C., Marz, J.,Robertson-Smith, B., & Vrba, A. (1993).TAPS—An O&M curriculum for students


with visual impairments. Austin: TexasSchool for the Blind and Visually Im-paired.

Sauerburger, D. (1989). To cross or not tocross: Objective timing methods of assess-ing street crossings without traffic controls.RE:view, 21, 153–161.

Sauerburger, D. (1995). Safety awareness forcrossing streets with no traffic control.Journal of Visual Impairment & Blindness,89, 423–431.

Sauerburger, D. (1999). Developing criteriaand judgment of safety for crossing streetswith gaps in traffic. Journal of Visual Im-pairment & Blindness, 93, 447–450.

Sauerburger, D. (2003, March 25). Do driversstop at unsignalized intersections for pe-destrians who are blind? Paper presented atthe Conference of Institute of Transporta-tion Engineers, Fort Lauderdale, FL.

Sauerburger, D. (2006). Instructional strate-gies for teaching judgment in detectinggaps for crossing streets with no trafficcontrols. RE:view, 37, 177–188.

Scheffers, W., & Myers, L. (1995). Inter-section analysis. In San Francisco StateUniversity orientation and mobilityreader (pp. L5–L8). San Francisco: SanFrancisco State University Departmentof Special Education.

Snook-Hill, M. M., & Sauerburger, D. (1996).Teaching students to assess safety forcrossing streets which have no trafficcontrol. In Proceedings of InternationalMobility Conference VIII (pp. 535–540).Melhus, Norway: Tambartun NationalResource Centre.

Wadhwa, L. C. (2003, January 16). Round-abouts and pedestrians with visual disabil-ities: How can we make them safer? Paperpresented at the Transportation ResearchBoard conference, Washington, DC.

Wiener, W. R., & Lawson, G. D. (1997).Audition for the traveler who is visuallyimpaired. In R. Welsh, W. R. Wiener, &B. B. Blasch (Eds.), Foundations of orien-tation and mobility (2nd ed., pp. 104–169).New York: AFB Press.

Wiener, W., Naghshineh, K., Salisbury, B., &Rozema, J. (2006). The impact of hybridvehicles on street crossings. RE:view, 38,65–78.

Wiener, W. R., & Siffermann, E. (1997). Thedevelopment of the profession of orienta-tion and mobility. In R. Welsh, W. R. Wie-ner, & B. B. Blasch (Eds.), Foundations oforientation and mobility (2nd ed., pp. 553–579). New York: AFB Press.

Wyle Laboratories. (1971, December 31).Community noise. Washington, DC: U.S.Environmental Protection Agency, Officeof Noise Abatement and Control. RetrievedJune 2, 2008, from http://www.nonoise.org/epa/roll14/roll14doc65.pdf

Robert Wall Emerson, Ph.D., associate professor,Department of Blindness and Low Vision, WesternMichigan University, 1903 West Michigan Ave-nue, Kalamazoo, MI 49008; e-mail: �[email protected]�. Dona Sauerburger, M.A.,COMS, orientation and mobility specialist, 1606Huntcliff Way, Gambrills, MD 21054; e-mail:�[email protected]�.


Documents

jjj