The impact of perceptual treatments on lateral control: A study using fixed-base and motion-base...

Accident Analysis and Prevention 40 (2008) 1513–1523

Contents lists available at ScienceDirect

Accident Analysis and Prevention

journa l homepage: www.e lsev ier .com/ locate /aap

Impact of perceptual treatments on lateral control during driving

Florence Roseya,∗, Jean-Michel Auberleta, Jean Bertrandb, Patrick Plainchault c

s – de

atal celate-off-rthe eon b

l; tha-basesectC), a

afterble s

here c

a INRETS, MSIS, 2 avenue du General Malleret-Joinville, F-94114 Arcueil cedex, Franceb CETE de l’Ouest, LRPC Angers, 23 Avenue de l’Amiral Chauvin, BP 69, F-49136 Les pontc CER-ESEO, 4 rue Merlet de la Boulaye, BP30926, F-49009 Angers Cedex 01, France

a r t i c l e i n f o

Article history:Received 12 November 2007Received in revised form 26 February 2008Accepted 27 March 2008

Keywords:Driving simulatorLane keepingUphillPerceptionCountermeasuresRoad safety

a b s t r a c t

Approximately 48% of all fcollisions. These crashes rFrance, single-vehicle run11%. This study evaluateddelineators, rumble stripsto maintain lateral controparticipants drove a fixedvertical curves (CVC). Foursection (i.e., the second CVsection (i.e., immediatelyof their lane with the rumwith the actual marking (

1. Introduction

In Europe, more than 80% of all fatal collision crashes occur-ring on rural roads, out of the urban context, are represented bythree accident types: single-vehicle (e.g., run-off-road and head-on collision with culverts and utility poles); head-on collisions; andcollisions at intersections. Single-vehicle run-off the road and head-on collisions (which relate to trajectory control), represent 48% ofall crash types (OECD, 1999) and inappropriate lateral positioningis one of the primary factors leading to crashes (RISER, 2006). Nev-ertheless, the topic of trajectory control, and more specifically the“lateral position” dimension and general vehicle path, has receivedlittle attention per se. Indeed, in very few studies has lateral positionbeen used as a central variable (e.g., Rasanen, 2005). The majorityof studies use lateral position variability as an indicator to evaluatecountermeasures concerning workload problems (e.g., Dukic et al.,2006; Reynaud et al., 2002; Rosenbloom, 2006; Sivak et al., 2006).

While human error is estimated to contribute to around 90%of crashes (Dewar and Olson, 2002; Wegman, 2007), road lay-

∗ Corresponding author. Tel.: +33 1 4043 6568; fax: +33 1 45 47 56 06.E-mail addresses: florence.rosey@wanadoo.fr (F. Rosey), auberlet@inrets.fr

(J.-M. Auberlet), jean.bertrand@equipement.gouv.fr (J. Bertrand),patrick.plainchault@eseo.fr (P. Plainchault).

– Ce Cedex, France

ollisions in Europe are classified as single-vehicle run-off-road or head-onto trajectory control (road departure) and represent a safety challenge. Inoad crashes represent 21% of all crashes and head-on collisions representffectiveness of four perceptual treatments (i.e., a painted center line, post-oth sides of the center line and sealed shoulders) in supporting the drivert is, to support the driver to keep in the center of his/her lane. Forty-threedriving simulator, on a simulated straight 3 km rural road with two crest

ions were chosen for analysis: a reference section (i.e., the first CVC), a testpre-test section (i.e., immediately before the second CVC) and a post-testthe second CVC). The results showed that drivers drive more at the centertrips on both sides of the center line and with the sealed shoulders thanenter line) or other treatments.

out has been identified as a contributing factor in about 30%(O’Cinneide, 1998; Rumar, 1985). Furthermore, a Road FederationBelgium report (2002) has shown that 20% of crashes are relatedto the road layout and 15% to road shoulders. Thus, it is often the

Forty-three participants with full French driving licenses (i.e.,not learners’ or restricted licenses) were recruited. The partici-pants were required to have a driving license for at least 2 yearsand normal or corrected-to-normal vision. One participant stoppedbecause of sickness discomfort, leaving 16 women and 26 menranging in age from 22 to 58 years (average age 38.6; S.D. = 10.83).Their average driving experience was 19 years, ranging from 5 to 35years, and they drove on average 12,373 km per year, ranging from5000 to 30,000 km. Fifty five percent of participants had more than15 years of driving experience. Upon their arrival in the laboratory,each participant was briefed on the requirements of the experimentand all read and signed an informed consent document.

2.2. Apparatus

The study was conducted using the INRETS-MSIS SIM2 drivingsimulator (Fig. 1), which is an interactive fixed-base driving sim-ulator with a complete Citroen Xantia car which hosts the user

1514 F. Rosey et al. / Accident Analysis

the edge of the road. It was found that variability in lateral posi-tion was reduced when a center line was added. Thus, the conceptsof driver expectancy and geometric consistency appear importantin safety and road design, because inconsistencies on a road cansurprise drivers and lead to errors that increase crash risk. Thisis understandable. First, road marking is the most important roadcharacteristic with respect to the recognition of road types that areused in a sustainably safe traffic system and, with respect to thespeed at which one is permitted to drive on different types of roads(Davidse et al., 2004). Second, the information provided by the roadand road environment is essential for the driver in order to mod-ulate driving control parameters and avoid risky behavior (Saad,2002; Theeuwes and Godthelp, 1995).

From previous sections, on the one hand, the situations at riskcould have a two-fold origin: infrastructure layout and/or the per-ception of the road environment and the infrastructure. On theother hand, the problems of trajectory control, more specially thoseof lateral position, appear well as a safety aim. Indeed, trajectorycontrol comprises a lateral dimension (i.e., vehicle position) and alongitudinal dimension (i.e., speed). Since 2000, in France, therehas been heavy enforcement of automated speed control. Thanksto this policy, a decrease in vehicle speed of 8 km/h (i.e., shift of90.7 km/h to 82.4 km/h) was observed in the period 2002–2006.This speed decrease contributed to a 75% reduction in deaths. Nev-ertheless, this tendency has slowed down since 2005 (SecuriteRoutiere, 2007). Consequently, the dimension of trajectory control(i.e., vehicle position) appears be interesting to explore, especiallygiven that the road safety consequences of changes in the lateralposition of road users (e.g., as a result of altered road markings) areless clear.

Relatively few studies have examined the impact of crest verti-cal curves (CVC) per se on driver performance. Usually they haveexamined the impact of CVC on driving performance and/or driverperception on horizontal curves (e.g., Bella, 2006; Bidulka et al.,2002; Hassan and Easa, 2003; Hassan and Sayed, 2002). Crest ver-tical curves interfere with trajectory control because as they reducesight distances; they hide the long-range visual information neededto predict the path of the road ahead and to anticipate future events(e.g., Rumar and Marsh, 1998) and to allow better lane keeping (e.g.,Summala, 1998). This absence of long-range information is all themore a threat since, during daytime, drivers stare frequently to thefar field than to the road edges (e.g., Serafin, 1994). Furthermore,with respect to vertical curves, design policy is based on the need toprovide drivers with adequate stopping sight distance (SSD). That

revention 40 (2008) 1513–1523

close to shoulders. These over-corrections, if the speed is too high,can lead to a fatal crash (e.g., collision or run-off-road).

The context for the conduct of the present study is theFrench national multidisciplinary research project PREDIT-SARI.This project aims to inform drivers and road managers more effec-tively of a heavy control loss risk on the rural network road. Morespecifically, our study concerns the risk related to the CVC onstraight rural roads.

The aim of the present study was to examine the impact of a rep-resentative range of perceptual countermeasures (PCM) on lateralcontrol; that is, which could allow the driver to better stay in thecenter of his/her lane, especially when driving on a CVC on a straightroad. The results would provide a better understanding of howdrivers deal with potentially useful guidance information that road-side elements (such as center line rumble strips, post-delineators. . .) provide with the acceptability level that is associated with theirpresence. The background problem was to provide additional visualcues about the road alignment and thus provide guidance informa-tion sufficiently in advance of any change in roadway heading, toallow the driver to plan and execute steering and speed controlmovements as smoothly as is needed for path maintenance.

2. Method

2.1. Participants

interface. The hardware is composed of four networked computers:one processes the motion equations and three generate the images.

The data recording system acquired all the objective parameters(e.g., relative position with respect to the road axis, local speed andaccelerations, steering wheel rotation angle, pedal actions, etc.) ata frequency of 60 Hz.

The simulator presented 3D driving scenes in panorama onto thethree screens (H: 150◦, V: 45◦, one in the center, and two on eachside) in a semicircular way. This set up provides a realistic view ofthe road and surrounding environment (Bella, 2005). The resolu-tion of the visual scene was 1024 × 768 pixels and the update ratewas 60 Hz. The simulator was placed at 2.8 m in front of the centralscreen from the participant’s head. The simulated road surface washigh friction, corresponding to dry asphalt, and scene visibility cor-responded to clear daytime conditions. The control devices werethe steering wheel, the manual transmission gears, and the pedals(i.e., brake, accelerator and clutch) of the complete Citroen Xantiacar. The driving simulator provided auditory feedback regarding carspeed, in the form of increased engine noise with increased speed,and haptic feedback with a force feedback steering wheel. For ourstudy, auditory feedback was provided when the simulated vehi-cle’s wheels crossed rumble strips on both sides of the center line(CRS) or sealed shoulders (SS). For these two treatments noise wasassociated to provide information in the case where a participantcould drive on. The scenario was updated dynamically accordingto the traveling conditions of the vehicle, which depended on thedrivers’ pedal and steering wheel control actions. Velocity was dis-

From the driver’s point of view, the simulator vehicle was usedand reacted to in the same way as a real car. Therefore, even to“start” the vehicle, the ignition key was used, resulting in enginenoises similar to a normal vehicle.

2.2.1. Simulator validityThe use of driving simulators presents a number of positive ele-

ments: experimental control, efficiency, low cost, safety, and ease ofdata collection (Bella, 2005, 2008). Nonetheless, we are aware thatsimulators must have appropriate validity in order to be regardedas a useful research tool. Two levels of validity are usually distin-guished (Bella, 2005; Kaptein et al., 1996): on the one hand, Absolutevalidity refers to the numerical correspondence between behaviorin the driving simulator and that in the real situation. On the otherhand, Relative validity refers to the correspondence between effectsof different variations in the driving situation. Relative validity isneeded for a driving simulator to be a useful research tool, but abso-lute validity is not essential (Tornos, 1998). As the research aim inour study was to deal with matters relating to the effects of inde-pendent variables (here, perceptual treatments and influence zone)

Fig. 2. Photograph (left) and reconstruction (right)

revention 40 (2008) 1513–1523 1515

and was not to determine numerical measurements (in referenceto Bella, 2005), we used a driving simulator.

More specifically, the aim of our study was to determine amongfour perceptual countermeasures (PCMs), which one had moreimpact on lateral position in the particular case of a CVC on astraight road. In other road contexts, these four PCMs have beenshown to be beneficial in an off-road test environment (Fildes andJarvis, 1994; Fildes and Lee, 1993; Mishra, 2006; Morena, 2003;Steyvers and de Waard, 2000). Consequently, because these fourPCMs have not been systematically examined in real CVC condi-tions, and in order to provide a safe, inexpensive and ethical facility,we used a driving simulator which is an ideal test environmentin which to address our issue (in reference to Godley et al., 1999;Kaptein et al., 1995, 1996). Recently, Keith et al. (2005) confirmedthe usefulness of driving simulators in the road-design process.

2.3. Simulator scenario

The roadway geometry depicted in the simulations was a recon-struction in virtual reality of the real rural two-lane road (D961) inMaine-et-Loire (Department 49, France). This reconstruction wasbased on the topographic layout, which consists of a straight roadwith two CVCs. The lane widths, road markings, sight distances,and other road engineering characteristics were incorporated intothe simulation in order to obtain similar road perception. Finally,

For the four experimental situations – the treatment control,post-delineators, rumble strips on both sides of the center linewhich named center line rumbles strips, and sealed shoulders –the delineation marking was the same before and after the testsection. For the “painted center line” situation there was no delin-eation marking before and after the section test. In the latter, thedelineation markings were deleted on all roads before and afterthe roadway section test. Furthermore, the longitudinal road profileand landscape were identical for all studied roads. Each treatmentwas applied only to one roadway section test which correspondedto 150 m immediately before and 150 m immediately after the toppoint of the second CVC. This roadway section test was named“test hill”. Studies suggested a preview time of 5 s for safe travel,especially when the sight distance is reduced (e.g., Godthelp andRiemersma, 1982; Rumar and Marsh, 1998; Weir and McRuer,1968). Based on this suggestion we chose test sections of 150 m.This 150 m distance corresponded with the distance traveled for

a theoretical speed of 90 km/h (i.e., maximum speed limit on therural road studied) during 6 s (i.e., 5 s + 1 s of tolerance).

In summary, the second CVC (i.e., roadway section test whichwas named “test hill”) is an experimental site identified by theproject SARI-VIZIR. While the BAAC (French police road accidentcasualty data) showed that there had been five fatal crashes onthe vertical crest curve on the site, during a visit to this site thefarmer’s interview, which took place close to the site, revealed that

2.4. Choice of experimental treatments

The choice of experimental treatments was realized in twostages. The first stage was a review of PCMs in the field andwith driving simulators. This showed that rumble strips, post-delineators, herringbone pavement markings, and so on, arerepresentative of PCMs and they had positive results on driverbehavior and/or in terms of crash reduction (e.g., Fildes and Jarvis,1994; Fildes and Lee, 1993; Fildes et al., 1997; Godley et al., 1999;Mishra, 2006; Morena, 2003; Steyvers and de Waard, 2000). Thesecond stage was a dialogue with road network engineers of theFrench Center for Technology Equipment Studies of NormandieCenter (CETE-NC) about possible PCMs. From this discussion fourPCMs (painted center line, post-delineators, rumble strips on both

Fig. 3. Pictures of the four experimental perceptual treatments as depicted in simulatiopoint of the second CVC (i.e., test hill). Picture at upper left: view of the post-delineator trfrom one delineator on the top point of the CVC. Picture at upper right: view of the centeline in order to be perceived by the participants. picture at lower left: view of the painteonly a painted center line on CVC. Picture at lower right: view of the sealed shoulders treabecause red and green are used for cycle paths in France. © 2007 MSIS INRETS.

revention 40 (2008) 1513–1523

sides of the center line and sealed shoulders, Fig. 3) were chosen forthe present study. It was the dialogue with road network engineersof CETE-NC that informed the final choice of the four PCMs. Indeed,the selection of PCMs for use on the real roads needed to takeaccount of not only their potential driving behavior benefits, butalso the practical realities that exist at the sites needing treatment –and the likely costs, benefits and maintenance issues. Furthermore,the use of the driving simulator allowed consideration, on the one

2.4.1. Painted center line (PC)We chose to place a painted center line only on the test hill –

i.e. without any delineation on the entire road before and after thetest hill. The reason for this choice was two-fold: first, because thedelineation serves as a reference for the driver to estimate boththe position and the speed of his own vehicle and other road users(Godthelp and Riemersma, 1982); and second, because delineationpermits the perception of lateral position. Indeed, the addition of acenter line has been found to minimize drivers’ variability in laneposition (Steyvers and de Waard, 2000; Triggs and Wisdom, 1979)and provide the main visual information in lane-keeping (Beall andLoomis, 1996; Beusmans, 1995; Riemersma, 1981, 1982). In the sim-ulator, studies dealing with the effect of markings have shown thatenhanced visibility of delineation leads to decreased lateral lane

n. Each perceptual treatment was implanted 150 m before and 150 m after the topeatment. The post-delineators were spaced 8 m apart. The implementation startedr line rumble strips. The rumble strips were implanted on both sides of the centerd center line treatment. The simulated road was without any marking, there was

tment. The width of the sealed shoulders was 1 m. The color ochre has been chosen

f delinRS waap of 1

F. Rosey et al. / Accident Analysis

Fig. 4. Examples of center line rumble strips (CRS). Picture A: raised CRS in side oimplantation in virtual scene. The length (perpendicular to the center line) of the C6 in.) and each CRS was implanted at 10 cm (i.e., 4 in.) from the center line, with a g

position errors (Horberry et al., 2006; McKnight et al., 1998). Fur-thermore, some studies have found that the presence of edge linesand a center line reduced all accidents by 20% (Miller, 1992) andsingle-vehicle accidents by 34% (Moses, 1986). The theoretical sen-sitivity for lateral position deviations may be better when the lateraldistance is smaller, and reaches an optimum when a guideline inthe center of the path is provided – i.e. the variability in lateral posi-tion decreases when the mean lateral distance to the delineationsystem is reduced (Godthelp and Riemersma, 1982).

In our study the painted center line was implemented only150 m immediately before and 150 m immediately after the toppoint of the second CVC (i.e., test hill), with no delineation outsidethe test hill.

2.4.2. Post-delineators (PoD)Post-delineators are light-reflecting devices mounted above the

roadway surface and along the roadway side in a series to indi-cate the roadway alignment and improve visual guidance. They area commonly used device for showing long continuous sections instraight as well as curved roads. Post-delineators are non-standarddevices for CVC but we chose them in our study to show the road-way vertical alignment in a guidance perspective and because theyare used in sections where changes in alignment might be confusingor unexpected.

In summary, the post-delineators were chosen instead of veg-etation (e.g., hedges) because of problems of contrast with thesimulated scene, i.e., as the simulated road was a rural road with

In our study, the post-delineators were implanted both 150 mbefore and 150 m after the top point of the second CVC (i.e., testhill). The post-delineators were spaced 8 m apart. This choice ofan 8 m interval derived from two French rules: (1) post-delineatorsshould be implemented with a minimal 5 m spacing and a maximal50 m spacing; and (2) the delineators must be spaced so that driverscan always see five post-delineators. In our case, with the CVC andin the driving simulator, five post-delineators were always visiblewith 8 m spacing.

2.4.3. Center line rumble strips (CRS)Rumble strips consist of either raised or grooved patterns,

installed perpendicular to the direction of travel, that produce audi-ble and tactile warnings when they are driven over by vehicle

revention 40 (2008) 1513–1523 1517

eation. Picture B: grooved CRS under delineation. Picture C: definition of the CRSs 27 cm (i.e., 10.63 in.), the width (along the center line) of the CRS was 15 cm (i.e.,0 cm between each rectangle.

tires. Center line rumble strips installed on shoulders can reducesingle-vehicle run-off-road crashes by approximately 20% (Griffith,1999; Hanley et al., 2000). Persaud et al. (2004) have also shownthat center line rumble strips are an effective countermeasure,on rural two-lane roads, and in preventing frontal and opposing-direction sideswipe crashes. The positive impact of the rumblestrips was effective under a variety of roadway geometric configu-rations (Persaud et al., 2004). The effect of center line rumble stripsis greater in enhancing safety than a painted line (Giaver et al., 1999in Bressan et al., 2003; Persaud et al., 2004), and they decrease lat-eral vehicle position away from center of the lane (Porter et al.,2004). These authors compared lateral vehicle placements beforeand after CRS implantation on four sites with two-lane rural roads.They showed that, for the 12 ft (i.e., 3.65 m) roads, the distancefrom the vehicle centroıd to the center line increased by 0.46 ft (i.e.,140 mm) and for the 11 ft (i.e., 3.35 m) roads the distance increasedby 0.25 ft (i.e., 80 mm) after the CRS implantation. Finally, Noyceand Elango (2004), with a driving simulator, found that they areboth effective at gaining drivers’ attention and a safety counter-measure in cross-over-the-center line fatal and injury crashes. Asprevious results have shown, this kind of device impacts more ondriver behavior than classic delineation, because the vibrations andthe noise which result from line crossings does not exist with apainted line (Rasanen, 2005).

In our study we have chosen to implement rumble strips on bothsides of the center line (CRS, Fig. 4) after dialogue with road networkengineers of CETE-NC and in order for the participants to perceivethis treatment. Each rectangle had a length (perpendicular to the

This treatment was both visual and auditory – i.e., when theparticipants drove on, the CRS produced a rumbling noise. We choseto present noise feedback only for the simulation of CRS devices,given that the fixed-base driving simulator used had no feedbackvibration system and that our focus of attention was mainly visualperception.

In our study, the CRS were implanted both 150 m before and150 m after the top point of the second CVC (i.e., test hill).

2.4.4. Sealed shoulders (SS)Generally, sealed shoulders are devices to increase usable

width and driver safety. In France, sealed shoulders are commonlyreserved for pedestrian paths, when they are implanted in a rural

of the vehicle’s centroıd was determined, as compared to the idealtrajectory, assumed to coincide with the driving lane axis (in refer-ence to Bella, 2005). A further reason for choosing lateral positionas a performance indicator was lane position variability, whichprovides a measure of driving performance that describes thesafety-relevance of changes in driving behavior (McGehee et al.,2004).

2.6.1.1. Lateral position. Lateral position is defined as the location ofthe vehicle’s longitudinal axis relative to a longitudinal road refer-ence system (Porter et al., 2004). For the purpose of our study, andin reference to Porter et al. (2004), the longitudinal road referencesystem used was the roadway center line, the roadway center delin-eation marking or the theoretical center-axis of the road betweenthe roadways in opposing directions when there was no delineationmarking. In our study, the lateral position corresponded to the dis-tance (in mm) of the vehicle centroıd to the roadway center line(Fig. 5). A lane position of 0 mm was obtained when the vehiclewas straddling the roadway center line.

Furthermore, four road sections (i.e., influence zones) were

1518 F. Rosey et al. / Accident Analysis

environment so cyclists can also use them. Surfacing the shoulderswith asphalt allows for the provision of visual, audible and tactileclues to drivers leaving the travel lane. In the present study thechoice of the color “ochre” for the texture was implemented afterdiscussion with road network engineers because green or red areused for cycle paths in France. The idea was that color variationbetween the shoulder and travel lanes adds visual delineation bydefining a pathway due to the color contrast, thus adding visualinformation to improve guidance.

This treatment was both visual and auditory – i.e., when theparticipants drove on, SS produced a passing noise to identify thepassage from lane to shoulder. This identification permits that thedriver to perform the recovery maneuver more quickly in the caseof leaving the roadway (CETE-NC, 1989).

To reiterate, the SS were chosen because this treatment is theobject, or has been the object, of field studies (CETE-NC, 1989; CETE-NC, 2002) and in France this device is uncommonly used as safetycountermeasure.

In our study, the sealed shoulders were implemented both 150 mbefore and 150 m after the top point of the second CVC (i.e., test hill).

2.5. Procedure

Each participant was tested individually. As a first step, upontheir arrival, the participants were briefed on the requirements ofthe experiment. All read and signed an informed consent docu-ment and were asked to complete a short questionnaire derivingbasic information (i.e., age, gender . . .). As a second step, the par-ticipants were introduced to the driving simulator and shown thefunctions of the vehicle. Then, they were familiarized with thesimulator in a neutral condition. In this condition, each partic-ipant drove on the reconstruction of a 3D virtual reality of theD961 rural road (but in the south direction, Segre to Marans) andhe/she had to take a roundabout, which allowed for familiariza-tion with the tactile feedback through the torque in the steeringwheel.

The participants were informed that they would drive on a ruralroad until they arrived at the roundabout and then they had to stopat the entry of the roundabout. The participants would adapt theirspeed to the driving conditions. In addition to a physical orientationto the apparatus and the expected task performance, participantswere instructed to “Please drive like you would drive in the samesituation in the real world, until the roundabout where you wouldstop at the entry”. The participants were informed that some dis-

Afterwards, the participants began with the reconstructed roadsection without any particular treatment, as it comprised onlythe delineation marking that currently exists (i.e., control road).Then, they drove through the experimental roadways on four sep-arate runs which were in random order. Each run consisted of onedifferent perceptual treatment. These perceptual treatments werepost-delineators, sealed shoulders, rumble strips on both sides ofthe center line, and centered line without edge side lines. All theparticipants drove in all situations: i.e., treatment control (refer-ence road); painted center line (PC); post-delineators (PoD); CRS;and SS.

For all driving conditions, there was a low concentration ofoncoming traffic from the contra flow lane to show the partici-pants that there could be cars. We chose not to create heavy trafficto avoid influencing participants’ speed and lateral positioning inreference to Lewis-Evans and Charlton (2006).

The participants finally drove five runs: one run consisted ofthe “real” (control) road and there was one run for each perceptualtreatment. A repeated measures design was used for the study.

revention 40 (2008) 1513–1523

After each driving run, the participant completed a short ques-tionnaire to gauge the acceptability level of the treatment (i.e.,comfort and safety of driving) immediately after the correspondingrun. These subjective data were used to estimate any potential dis-crepancies between the participant feelings and the objective datarecorded.

Finally, after the five runs, participants were asked to completetwo questionnaires, one concerning the simulation realism (i.e.,scene visualization and driving simulator) and another concerningtheir driving experience – i.e., license years, number of kilometers,accidents . . .

The experiment took an average of 1 h 30 min for participants tocomplete; the driving phase represented an average of 45 min.

2.6. Data collection and Statistics Analysis

2.6.1. Data collectionAmong the number of parameters collected by the simulator

we processed lateral position. The lateral position measurements(i.e., lateral position) were continuously recorded with a samplingfrequency of 60 Hz.

To study driver’s behavior concerning the trajectory adoptedto drive along the crest vertical curve according to the perceptualtreatment, the reference lateral position of the lateral placement

determined (Fig. 6): the pre-test hill section (PrTH) corresponded to

Fig. 5. Example of method used to determine the lateral position. The lateral posi-tion corresponded to the distance (in mm) of the vehicle centroıd (empty silhouette(�)) to the roadway center line (CR, which corresponded to the longitudinal roadreference system, thus equal to 0). A lane position of 0 mm was obtained when thevehicle was straddling the roadway center line.

ns stuhe fou

posim of t

Fig. 6. View of the profile of the 3D reconstructed road with the four test hill sectioThe first CVC hid the second thus the drivers could not pre-view the second CVC. Tfirst CVC and was determined as section of reference to identify the lateral vehicleto the 150 m before treatment. Test hill section (TH, red) corresponded to the 300

the 150 m before treatment of the second CVC; the test hill section(TH) corresponded to the 300 m of treatment on the second CVC(i.e., 150 m on each side of the top point of the CVC); and the post-test section (PoTH) corresponded to the 150 m after treatment. Afourth section was determined, as a section of reference, to identifythe lateral vehicle position preferred beyond any treatment. Thissection of reference (SR) corresponded to the 150 m before the firstCVC before the second CVC (Fig. 4). These four sections of road weredetermined to identify the existence of treatment impact accord-ing to the time of perception; i.e., before (pre-test hill), during (testhill) and after (post-test hill).

2.6.2. Statistic analysisLateral position was measured repeatedly for all participants

across the five perceptual treatment conditions. Then, lateral posi-tion measures were analyzed using repeated measures ANOVAwith perceptual treatment (treatment control, painted centerline, post-delineators, rumble strips on both sides of the centerline, and sealed shoulders) and influence zone (section of refer-

Table 1Summary of the means and standard deviations of lateral position (mm) with respect to t

Roadway sections Treatments Lateral position (mm)

Mean Standard deviation

Section of reference TC 1394 215PC 1509 275PoD 1540 217CRS 1485 179SS 1530 202

Pre-test hill TC 1407 224PC 1546 264PoD 1524 182CRS 1520 178SS 1576 177

Test hill TC 1472 189PC 1541 181PoD 1488 123CRS 1630 151SS 1619 150

Post-test hill TC 1466 238PC 1561 257PoD 1515 179CRS 1602 209SS 1604 178

Summary of the difference means in the TC means and treatment means, and in the Sn andthe road center line. Note: TC, treatment control; PC, painted center line; PoD, post-deline

a Difference in the TC mean and Treatment mean for each roadway section. If the diffePCM installation from the delineation marking of reference.

b Difference in the Sn roadway section mean and Sn+1 roadway section mean. If the di(right) away from the lateral position on Sn .

revention 40 (2008) 1513–1523 1519

died. The straight road was composed to two CVC, the CVC studied was the second.r sections were: small hill section (SH, green) corresponded to a 150 m portion on

tion preferred out all treatment. Pre-test hill section (PrTH, yellow) correspondedreatment on second CVC (i.e., on each 150 m side of the top point of the CVC) and

ence, pre-test hill, test hill, and post-test hill) as within-subjectsfactors.

Before performing repeated measures ANOVAs, all data used inthe statistical analysis were submitted to the Kolmogorov–Smirnovtest, which showed that all data were normally distributed. Thistest was not sufficient, however, because the F test is robustto violations of the multivariate normal assumption, but notto the sphericity assumption (Lewis, 1993). When the spheric-ity assumption is violated (i.e., here when the Mauchley testwas significant) adjustments were made to the ANOVA resultsusing the Geisser–Greenhouse epsilon, which provides an F-testusing a much more stringent criterion (in reference to Geisserand Greenhouse, 1958). Thus, the decision about whether an F-test was significant was made based on the Geisser-Greenhouseepsilon. Where the within-subject variables violated the sphericityassumption, we have reported Geisser–Greenhouse probabili-ties. Each repeated measures ANOVA was followed by post hocNewman–Keuls test. The threshold for statistical significance wasset at .05.

he five perceptual treatment conditions and the four roadway sections

Magnitude of change (mm) Treatments Magnitude of change (mm)

Difference TC-treatmenta Difference Sn − Sn + 1b

TC 12114 65145 −690

135 PC 36

−5138 20117113 PoD −16169 −36

276816 CRS 35

158 110147 −28

94 SS 4748 43

136 −15138

Sn+1. The lateral position corresponded to the distance of the vehicle centroıd fromators; CRS, rumble strips on both sides of center line, SS, sealed shoulders.rence is not equal to 0, change in the mean lateral position occurred as a result of

fference is negative (positive) then the treatment on Sn+1 lead to a shift to the left

3. Results

For the sake of clarity, it should be noted that, for lateral position,the reference point was the center axis of the road. Table 1 presentsthe means and standard deviations of lateral position (mm) withrespect to the five perceptual treatment conditions and the fourroadway sections. This table presents also the difference in meansbetween the treatment control and each treatment with respect tothe four roadway sections and, the difference in means betweena roadway section (Sn) and the roadway immediately after (Sn+1)with respect to the five perceptual treatment conditions.

3.1. The effects of perceptual treatments

The repeated measures ANOVA computed on the effects ofperceptual treatments indicated a reliable perceptual treatmenteffect for the pre-test hill (PrTH, F(3.10,127.25) = 4.57, p < .001, thetest hill (TH, F(4,164) = 8.76, p < .0001) and the post-test hill (PoTH,F(4,164) = 2.99, p < .02) – but not for the section of reference (SR,F(3.32,136.16) = 2.10, p = .09). Thus, the participants drove, on thesmall hill which hid the test hill, with the same trajectory meanfor the reference road (i.e., actual simulated road) and the treatedroadway.

Post hoc analyzes indicated that:

(1) for the pre-test hill, the participants drove on the reference road(RR) closer to the center of the road than on the treated roads(RR*PC, p < .006; RR*PoD, p < .01; RR*CRS, p < .007 and RR*SS,p < .0006). Furthermore, they indicated that the participantsdrove with similar trajectories for all the perceptual treatments.

(2) for the test hill, participants drove on the RR, on the PC and onthe PoD closer to the center of the road than on the rumble stripson both sides of the center line (CRS*RR, p < .0001; CRS*PC,p < .02; CRS*PoD, p < .0001) and the sealed shoulders (SS*RR,p < .0001; SS*PC, p < .0001; SS*PoD, p < .0003). Other than thatthey drove with similar trajectories on RR and PC (p = .12), RRand PoD (p = .75), on PC and PoD (p = .13) and, on CRS and SS(p = .74).

(3) for the post-test hill, the participants drove on the RR closer tothe center of the road than on the road with CRS (p < .02) andSS (p < .03).

In summary, the participants drove according to two types oftrajectories: on the one hand, there were those who drove closer

3.2. Influence zones (sections)

The repeated measures ANOVA computed on the effects ofinfluence zone indicated a reliable influence zone effect for thereference road (F(2.53,131.36) = 4.47, p < .007), the rumble strips onboth sides of the center line (F(2.32,95.36) = 14.50, p < .0001) andthe sealed shoulders (F(2.30,94.24) = 4.10, p < .01); but not for thepainted center line treatment (F(2.33,95.37) = .46, p = .66) or thepost-delineators treatment (F(2.12,89.28) = 1.09, p = .34). Thus, theparticipants drove with similar trajectories for the painted cen-

revention 40 (2008) 1513–1523

ter line and post-delineators treatments whatever the zone (i.e.,before, during, and after).

Post hoc analyzes indicated for the RR and the rumble strips onboth sides of the center line (CRS) that:

(1) the participants drove with similar trajectories on the sectionof reference (i.e., 150 m before the first CVC and pre-test hill(respectively, RR, p = .64 and CRS, p = .16), and also on the testhill and on the post-test hill (respectively, RR, p = .83 and CRS,p = .26).

(2) the participants drove closer to the center of the road on thepre-test hill (i.e., before the treatment) than on the test hill (i.e.,treated CVC, respectively, RR, p < .03 and CRS, p < .001) and thanon the post-test hill (i.e., after the treatment, respectively, RR,p < .02 and CRS, p < .001).

The post hoc analyzes indicated for the sealed shoulders that theparticipants drove with a similar trajectory on the 150 m before thetreated CVC (PrTH), on the treated CVC (TH) and on the 150 m afterthe treated CVC (PrTH) (PrTH*TH, p = .44; PrTH*PoTH, p = .75 andTH*PoTH, p = .36).

In summary, the results from the study of influence zone showedthat rumble strips on both sides of the center line (CRS) and SS areeffective to focus the drivers on the lane both immediately afterentering the treatment CVC zone as well as throughout the treat-ment CVC zone. Furthermore, the results show that the treatmentcontrol (i.e., actual delineation marking, painted center line anddelineators) had little effect on lateral position relative to the CRSand SS treatments.

4. Discussion

To keep in his/her lane, the driver uses mainly visual perceptionwhich is based on elements of the driving environment such as roadmarkings, road delineation, etc. . .. From this assumption, relativelyto our focus of attention that was the trajectory control according tothe perception theory, we have analyzed the impact of four percep-tual treatments on drivers’ trajectory especially during simulateddriving on straight roads with two CVC. In our particular context(i.e., crest vertical curve), the study examined whether PCMs, couldimpact drivers’ lateral positions and how they impacted them whenthey drove on treated CVC with these PCMs. The analysis of lat-eral position according to the perceptual treatments yielded two

4.1. Perceptual treatments and trajectory control

Our results showed that, with the presence of PCMs on CVC –i.e., here, painted center line, post-delineators, rumble strips onboth sides of the center line or sealed shoulders – the mean lateralposition change was further to the right from the roadway centerline compared with the actual simulated road (i.e., reference roadwith actual delineation marking, shoulders . . .). On the referenceroad, the distance from the vehicle centroıd to the roadway centerline was on average 1435 mm (S.D. = 39.73); in the PCM sections,the distance from the vehicle centroıd to the roadway center lineincreased on average to 114 mm (S.D. = 42). These findings suggestthat in the case where a vehicle is coming in a contra flow laneduring the CVC crossing, the two vehicles would tend to pass with

greater margin, since the perceptual treatment incites the driversto adopt a trajectory further to the right of the roadway center line,thus closer to center of the travel lane.

More specifically, the results have also shown two categoriesof trajectories: a “close to roadway center line” trajectory – i.e.,when the shift from the roadway center line was inferior to theshift mean (i.e., 114 mm) – and an “away from the roadway centerline” trajectory – i.e., when the shift from the roadway center linewas superior to the shift mean (i.e., 114 mm). The “close to road-way center line” trajectories were observed for the reference road(RR, 1435 mm), the painted center line (+104 mm/RR) and the post-delineators treatments (+81 mm/RR). The “away from the roadwaycenter line” trajectories were observed for the rumble strips on bothsides of the center line (+124 mm/RR) and sealed shoulders treat-ments (+147 mm/RR). These results show that the rumble stripson both sides of the center line (CRS) and the SS impact furtherthe trajectory control by a shift away from the roadway center linenearly equivalent to a tire width. The results obtained for the CRSand SS are in the same direction as those obtained by Porter et al.(2004), in a study comparing lateral position before and after CRSimplantation on field sites. In their study the authors found a shiftaway of 0.46 ft (i.e., 140 mm) and 0.25 ft (i.e., 76 mm) of the lat-eral position mean after CRS implantation compared with beforeCRS implantation. Furthermore, the positive influence of the CRSon drivers’ behavior is comparable to Noyce and Alango’s results(2004) concerning this treatment with driving simulator analysis.Those results concerning the sealed shoulders are comparable tothose of the CETE-NC’s study (2002) on an actual site (road DR 982,Seine-Maritime, France). This study showed that the drivers drovecloser to the right side of the pavement with sealed shoulder instal-lation. In our study we have found that the sealed shoulders have asimilar impact to that of the rumble strips on both sides of the cen-ter line on trajectory control; that is to say, the driver drove furtherto the right of the roadway center line. The SS treatment is usefulin decreasing rollover crashes and its advantage is that drivers havean additional security margin to regain control in the case of lossof vehicle control, relative to borderline rumble strips. The otheradvantage is that, thanks to the contrast between road edge androad shoulder provided by sealed shoulders, the effort to keep thevehicle on the road is decreased and consequently the probabilityof overcorrection by the drivers which may lead them to end upon the other side of the road would decrease. Data suggest thatedge-lines counteract the increasing of effort to keep a vehicle onthe road when there is a low contrast between road edge and road

Furthermore, rumble strips on both sides of the CRS or SS appearto be effective trajectory control devices and safety measures inthe case of CVC. These two perceptual treatments led the drivers todrive close to center the lane, which may help to prevent eventualcollisions with vehicles in contra flow. In the case of a small roadwidth (i.e., 3 m), these eventual collisions could result from drift-ing, swerving or straddling the center line while driving on CVCconsequently to a placement of the vehicle closer to the center-line. Furthermore, as the SS on lateral position was similar to thatfor CRS, thus they appears as an alternative devices to CRS, whichcorresponds to the recommendations of Noyce and Elango (2004).

The fact that the trajectories are not significantly influenced bypost-delineators treatment whatever the section of road (p = .10 top = .74), might suggest that the drivers used more horizontal mark-ings than vertical. Another assumption, from the results on the

revention 40 (2008) 1513–1523 1521

differences of lateral position between reference road and post-delineator road results, is that the driver could be stressed bythis treatment and motivated temporarily to “escape”; to moveaway from that obstacle. This behavior (i.e., left away from post-delineators) is similar to that experienced when we drive in a tunnelfor example.

4.2. Influence zones of the perceptual treatments and trajectorycontrol

The results have shown that there are influence zones, i.e. thatthe perceptual treatments have an impact not only when the driverdrove on the treatment area (i.e., the second CVC) and that to theimpacts depend on the perceptual treatment.

There are no influence zones for the painted center line orpost-delineators treatments, since the trajectories were similarwhatever the zone (i.e., before, during, after). But, the CRS impacteddrivers’ trajectory on the treatment zone and on the 150 m after thetreatment zone, whereas the sealed shoulders impacted drivers’trajectory before, during and after the treatment zone. From theseresults, we suggest that the sealed shoulders could allow forsmoother steering in the sense that before driving on the CVC thedrivers adopted a lateral position away from the roadway centerline. The fact that the rumble strips on both sides of the center linehad a lesser impact on the 150 m before the treated CVC can beexplained by the result that 30% of drivers have not seen this treat-ment. Whereas the rumble strips on both sides of the center lineare not a common device in France and they are of low contrast,the drivers appeared influenced when they drove on the CVC, i.e.,where the rumble strips on both sides of the center line have beenimplanted.

To conclude, the rumble strips on both sides of the center lineand the sealed shoulders impacted the driver’s lateral position by amove away from the roadway center line, which was approximatelyequivalent to one tire width. These two perceptual treatmentsappear to be effective devices in maintaining trajectory controlin the case of CVC and in decreasing lateral drifts and, by conse-quence decreasing the probability of crossing or overlapping of thecenter line when the driver is crossing the CVC. Their impact ondrivers’ lateral position in the treatment zone is consistent witha perceptual effect. Here, in the CVC context, they appear to havewell helped the drivers to make up the absent perceptual infor-mation (i.e., road geometry and road delineation after the CVC) tocontrol their trajectory. We are aware that without the speed data

4.3. Recommendations for further research

The results of this study must be taken with caution concern-ing the generalization from a driving simulator environment inthe laboratory to a real-life driving environment. Simulator stud-

ies provide a suitable and controllable environment; nevertheless,the driving simulator is not a total substitute for the real worldexperiences afforded by on-road driving. Thus, experiments on realroads are needed to firmly establish the benefits on the road itself.An emerging consensus, however, is that relative validity may bemore important than absolute validity for a driving simulator to bea useful research tool (Godley et al., 2002). Our results have shownthat it is the rumble strips on both sides of the center line andthe sealed shoulders that have the greater impact on the lateralposition by right shift toward the center lane. From these resultsthe general council of Maine-et-Loire (Department 49, France) willexperiment with one of these two treatments, and may be both ofthem, at a real site. The experiment on a real site will allow for fur-ther evaluation of lane position benefits. Driving simulators havefurther benefits as a research tool, in the sense that they allowresearchers to pre-determine, in a safe environment, treatmentsfor use on public roads within risky contexts such as crest verti-cal curves, horizontal curves and intersections with degraded sightdistances. Use of the simulator is all the more interesting in thecase of uncommon technical devices as in our study; that is, forthe use of rumble strips on both sides of the center line which arenot used in France. Finally, the evidence that the rumble strips onboth sides of the center line impacted the drivers’ trajectory eventhough 30% of drivers had not previously seen them, suggests thata treatment can implicitly impact driver behavior. The use of a driv-ing simulator with subjective and objective (i.e., collected) data canallow researchers to anticipate in advance both the desirable andunwanted effects of the treatment.

Acknowledgments

The research presented in this paper was performed as part ofthe French National project SARI-VIZIR. We thank the anonymousreviewers for helpful comments and valuable guidance on an ear-lier version of the manuscript and Michael Regan for her helpfulcomments. We thank Olivier Menacer of SETRA, Guy Bertrand andOlivier Moisan of CETE Normandie-Center for their support andhelpful comments, Marie-Pierre Pacaux-Lemoine, Sabrina Huvelleand Francoise Anceaux of the LAMIH for the questionnaire elabora-tion. We thank also Isabelle Aillerie and Fabrice Vienne of the FrenchNational Institute for Transport and Safety Research (INRETS-MSIS)for the 3D-Base, Sebastien Boutelier of INRETS-CIR for the data“treatment” and, Ariane Tom and Guillaume Savaton for their help-

References

Beall, A.C., Loomis, J.M., 1996. Visual control of steering without course information.Perception 25, 481–494.

Bella, F., 2005. Driving simulator validation for work-zone design. Transport. Res.Rec. 1937, 136–144.

Bella, F., 2006. Effects of driver perception of combined curves on speed and lateralplacement. In: TRB 85th Annual Meeting Compendium of Papers CD-ROM.

Bella, F., 2008. Driving simulator for speed research on two-lane rural roads. Accid.Anal. Prev. 40, 1078–1087.

Beusmans, J., 1995. Center of Outflow is Not Used to Control locomotion (Rep. CBRTR No. 95-5). Nissan Cambridge Basic Research Laboratory, Cambridge, MA.

Bidulka, S., Sayed, T., Hassan, Y., 2002. Influence of vertical alignment on horizontalcurve perception: phase I examining the hypothesis. Transport. Res. Rec. 1796,12–23.

Blana, E., Golias, J., 2002. Differences between vehicle lateral displacement on theroad and in a fixed-base simulator. Hum. Factors 44, 303–313.

Bressan, P., Garlaschelli, L., Barracano, M., 2003. Antigravity hills are visual illusions.Psychol. Sci. 14, 441–449.

CETE-NC, 1989. Etude de l’influence des accotements revetus sur le comportementdes usagers. CETE Normandie Center, novembre 1989.

CETE-NC, 2002. Influence de la realisation d’accotements revetus sur le comporte-ment des usagers: Cas de la RD982 en Seine-Maritime. CETE Normandie-Center,Novembre 2002.

revention 40 (2008) 1513–1523

COST, 1999. Requirements for horizontal road markings (COST Project 331). Officefor Official Publications of the European Communities, Luxembourg.

Claassen, A., Pouwels, B., 1998. Visuele informatie als stuurcomponent voor gewenstrijgedrag; Een literatuurstudie (Visual information as steering component fordesired driving behavior: A literature review). Institute for Applied Social Sci-ences (ITS), Nijmegen University, Nijmegen (in Dutch).

Davidse, R.J., van Driel, C.J.G., Goldenbeld, Ch., 2004. The effect of altered road mark-ings on speed and lateral position: a meta-analysis. R-2003-31. SWOV.

De Waard, D., Steyvers, F.J.J.M., Brookhuis, K.A., 2004. How much visual road infor-mation is needed to drive safely and comfortably? Safety Sci. 42, 639–655.

Dewar, R.E., Olson, R., 2002. Human Factors in Traffic Safety. Lawyers and JudgesPublishing, Tucson, AZ.

Donges, E., 1978. A two-level model of driver steering behavior. Hum. Factors 29,691–707.

Dukic, T., Hanson, L., Falkmer, T., 2006. Effect of drivers’ age and push button loca-tions on visual time off road, steering wheel deviation and safety perception.Ergonomics 15 (49), 78–92.

Fildes, B.N., Godley, S.T., Triggs, T.J., Jarvis, J., 1997. Perceptual Countermeasures:Simulator Validation Study (CR 169 (FORS)/RR 1/97 (RTA)). Roads and TrafficAuthority of N.S.W. and Federal Office of Road Safety, Sydney/Canberra, Australia.

Fildes, B.N., Jarvis, J.R., 1994. Perceptual Countermeasures: Literature Review (Res.Rep. CR4/94). Roads and Traffic Authority of N.S.W. and Federal Office of RoadSafety, Sydney/Canberra, Australia.

Fildes, B.N., Lee, S.J., 1993. The Speed Review: Road Environment, Behavior, SpeedLimits, Enforcement and Crashes (Rep. Nos. CR 127 (FORS) CR 3/93 (RSB)). Fed-eral Office of Road Safety; Road Safety Bureau, Roads and Traffic Authority,Sydney/Canberra, Australia.

Geisser, S., Greenhouse, S.W., 1958. An extension of Box = s results on the use of theF distribution in multivariate analysis. Ann. Math. Stat. 29, 885–891.

Giaver, T., Sakshaug, K., Jenssen, G.D., Berge, T., 1999. Tiltak for reduksjon av strekn-ingsulykker. Delrapport 2. Effekter av profilert vegmerkning (STF22 A99553).SINTEF, Trondheim, Bygg, miljoteknik, sam ferdsel.

Godley, S.T., Triggs, T.J., Fildes, B.N., 2002. Driving simulator validation for speedresearch. Accid. Anal. Prev. 24, 589–600.

Godley, S.T., Fildes, B.N., Triggs, T.J., Brown, L.J., 1999. Perceptual Countermeasures:Experimental Research (CR 182). Australian Transport Safety Bureau (ATSB),Canberra, Australia.

Godthelp, H., Riemersma, J.B.J., 1982. Perception of Delineation Devices in Road WorkZones During Nighttime (Rep. No. 820413). Society of Automotive Engineers,New York, NY.

Griffith, M.S., 1999. Safety evaluation of rolled-in continuous shoulder rumble stripsinstalled on freeways. Transport. Res. Rec. 1665, 28–35.

Hanley, K.E., Gibby, A.R., Ferrara, T.C., 2000. Analysis of accident-reduction factorson California state highways. Transport. Res. Rec. 1717, 37–45.

Harms, L., 1993. Driving performance on a real road and in a driving simulator: resultsof a validation study. In: Presented at 5th Conference on Vision in Vehicles.Glasgow, Scotland.

Hassan, Y., Easa, S.M., 2003. Effect of vertical alignment on driver perception ofhorizontal curves. J. Transport. Eng. 129, 399–407.

Hassan, Y., Sayed, T., 2002. Effect of driver and road characteristics on requiredpreview sight distance. Can. J. Civil Eng. 29, 276–288.

Horberry, T., Anderson, J., Regan, M., 2006. The possible safety benefits of enhancedroad markings: a driving simulator evaluation. Transport. Res. F Traffic Psychol.Behav. 9, 77–87.

Johnston, I.R., 1983. The Effects of Roadway Delineation on Curve Negotiation byBoth Sober and Drinking Drivers (ARR 128). Australian Road Research Board,Victoria.

Kallberg, V.P., 1993. Reflector posts—signs of danger? Transport. Res. Rec. 1403,57–66.

Kaptein, N.A., Theeuwes, J., van der Horst, A.R.A., 1996. Driving simulator validity:some considerations. Transport. Res. Rec. 1550, 30–36.

Kaptein, N.A., Theeuwes, J., van der Horst, A.R.A., Van Velze, R., 1995. Behavioralevaluation of tunnel design characteristics. In: Proceedings of Driving SimulatorConference, Sophia Antipolis, France, pp. 409–420.

Keith, K., Trentacoste, M., Depue, L., Granda, T., Huckaby E., Ibarguen, B., Kantowitz,B., Lum, W., Wilson, T., 2005. Roadway human factors and behavioral safety inEurope. Report No.: FHWA-PL-05-005.

Land, M.F., Lee, D.N., 1994. Where we look when we steer? Nature 369, 742–744.Land, M.F., Horwood, J., 1995. Which parts of the road guide steering? Nature 377,

339–340.Lewis, C., 1993. Analyzing means from repeated measures data. In: Keren, G., Lewis,

C. (Eds.), A Handbook for Data Analysis in the Behavioral Sciences. Erlbaum,Hillsdale, NJ, pp. 73–94.

Lewis-Evans, B., Charlton, S.G., 2006. Explicit and implicit processes in behavioraladaptation to road width. Accid. Anal. Prev. 38, 610–617.

McGehee, D.V., Lee, J.D., Rizzo, M., Dawson, J., Bateman, K., 2004. Quantitative anal-ysis of steering adaptation on a high performance fixed-base driving simulator.Transport. Res. F Traffic Psychol. Behav. 7, 181–196.

McKnight, A.S., McKnight, A.J., Tippetts, A.S., 1998. The effect of lane line width andcontrast upon lane keeping. Accid. Anal. Prev. 30, 617–624.

Miller, T.R., 1992. Benefit-cost analysis of lane marking. Transport. Res. Rec. 1334,38–45.

Mishra, A., 2006. Analysis of run-off-the road crashes involving overcorrection: astudy of fatal traffic crashes in Florida in 2000. Unpublished M.Sc. Thesis. FloridaState University-FAMU-FSU College of Engineering, Florida, USA.

Summala, H., 1998. Forced peripheral vision driving paradigm: evidence for the

Morena, D., 2003. The nature and severity of drift-off road crashes on Michiganfreeways, and the effectiveness of various shoulder rumble strip designs. In:Transportation Record Board, annual meeting.

Moses, P.J., 1986. Edge lines and single vehicle crashes. Western Roads, pp. 6–8.Mutabazi, M.I., Russell, E.R., Stokes, R.W., 1998. Driver’s attitudes, understand-

ing, and acceptance of passing lanes in Kansas. Transport. Res. Rec. 1628,25–33.

Noyce, D.A., Elango, V.V., 2004. Safety evaluation of centerline rumble strips. Trans-port. Res. Rec. 1862, 44–53.

O’Cinneide, D., 1998. The relationship between geometric design standards andsafety. International symposium on highway geometric design practices, 44, pp.1–7.

OECD, 1999. Safety strategies for rural roads. Road transport and intermodal researchreport. Organisation for economic co-operation and development, Paris.

Persaud, B.N., Retting, R.A., Lyon, C.A., 2004. Crash reduction following installa-tion of centerline rumble strips on rural two-lane roads. Accid. Anal. Prev. 36,1073–1079.

Porter, R.J., Donnell, T.E., Mahoney, K.M., 2004. Evaluation of effects of centerlinerumble strips on lateral vehicle placement and speed. Transport. Res. Rec. 1862,10–16.

Ranney, T., Gawron, V., 1986. The effects of pavement edgelines on performance ina driving simulator under sober and alcohol-dosed conditions. Hum. Factors 28,511–525.

Rasanen, M., 2005. Effects of a rumble strip barrier line on lane keeping in a curve.

Reynaud, M., Le Breton, P., Gilot, B., Vervialle, F., Falissard, B., 2002. Alcohol is themain factor in excess traffic accident fatalities in France. Alcohol. Clin. Exp. Res.26, 1833–1839.

Riemersma, J.B.J., 1981. Visual control during straight road driving. Acta Psychol. 48,215–225.

Riemersma, J.B.J., 1982. Perception and Control of Deviations From a Straight Course:A Field Experiment (Rep. TNO-IZF 1982-C20). TNO Human Factors ResearchInstitute, Soesterberg, Netherlands.

RISER, 2006. European best practice for roadside design: Guidelines for roadsideinfrastructure on new and existing roads. RISER Deliverable D06. Goteburg,Chalmers University of Technology, 2006.

Road Federation Belgium, 2002. Chaınons Manquants et TravauxRoutiers Prioritaires. Conference de Presse du 23 octobre 2002.http://www.rfbelgium.be/press/broch fr.pdf.

Rosenbloom, T., 2006. Related Articles, driving performance while using cell phones:an observational study. J. Safety Res. 37, 207–212.

Rumar, K., 1985. The role of perceptual and cognitive filters in observed behavior. In:Evans, L., Schwing, R.C. (Eds.), Human Behavior and Traffic Safety. Plenum Press,New York, pp. 151–165.

Rumar, K., Marsh, D.K., 1998. Lane Markings in Night Driving: A Review of PastResearch and of the Present Situation (Technical Report No. UMTRI-98-50). TheUniversity of Michigan Transportation Research Institute, Ann Arbor.

Russell, E.R., Rys, M.J., 2005. NCHRP Synthesis 339: Centerline Rumble Strips. Trans-portation Research Board, Washington, DC.

Saad, F., 2002. Ergonomics of the driver’s interface with the road environment: thecontribution of psychological research. In: Fuller, R., Santos, J.A. (Eds.), HumanFactors for Highway Engineers. Pergamon, Oxford, pp. 23–41.

revention 40 (2008) 1513–1523 1523

Schieber, F., 2000. Highway research to enhance the safety and mobility of olderroad users. In: Conference Proceedings 27: Transportation in an Aging Society: ADecade of Experience. Transportation Research Board of the National Academies,Washington, DC, pp. 125–154.

Securite Routiere, 2007. Bilan de l’accidentologie de l’annee 2006 : comporte-ment. Resultats synthetiques vitesse et ceinture. http://www.securiteroutiere.equipement.gouv.fr/IMG/Synthese/CT SYNT.pdf.

Serafin, C., 1994. Driver Eye Fixations on Rural Roads: Insight Into Safe DrivingBehavior (Technical Report No. UMTRI-94-21). The University of Michigan Trans-portation Research Institute, Ann Arbor.

Sivak, M., Schoettle, B., Reed, M.P., Flannagan, M.J., 2006. Influence of visibility outof the vehicle cabin on lane-change crashes. Accid. Anal. Prev. 38, 969–972.

Smith, B.L., Lamm, R., 1994. Coordination of horizontal and vertical alignment withregard to highway aesthetics. Transport. Res. Rec. 1445, 73–85.

Staplin, L., Lococo, K., Byington, S., Harkey, D., 2001. Guidelines and recom-mendations to accommodate older drivers and pedestrians. Report No.:FHWA-RD-01-051.

Steyvers, F.J.J.M., 1994. Effecten van wegbelijning op rijgedrag en verkeersveiligheid:een literatuuroverzicht (Effects of road markings on driving behavior and roadsafety: a literature review), Report WR 94-05. Traffic Research Center, Universityof Groningen, Haren (in Dutch).

Steyvers, F.J.J.M., de Waard, D., 2000. Road-edge delineation in rural areas: effectson driving behavior. Ergonomics 43, 223–238.

hypothesis that car drivers learn to keep in lane with peripheral vision. In: Gale,A., Brown, I.D., Haslegrave, C.M., Taylor, S.P. (Eds.), Vision in Vehicles VI. Elsevier,Amsterdam, pp. 51–60.

Texas Department of Transportation, 2006. Texas Roadway Design Manual (October2006, Revised). ftp://ftp.dot.state.tx.us/pub/txdot-info/gsd/manuals/rdw.pdf.

Theeuwes, J., Godthelp, H., 1995. Self-explaining roads. Safety Sci. 19, 217–225.Tornos, J., 1998. Driving behavior in a real and a simulated road-tunnel: a validation

study. Accid. Anal. Prev. 30, 497–503.Triggs, T.J., Fildes, B.N., 1986. Roadway delineation at night. In: Gale, A.G., et, al. (Eds.),

Vision in Vehicles. Elsevier, Amsterdam, pp. 375–384.Triggs, T.J., Wisdom, P.H., 1979. Effects of Pavement Delineation Marking on Vehicle

Lateral Position Keeping. Department of Psychology, Monash University, Clayton,Victoria.

van Driel, C.J.G., Davidse, R.J., van Maarseveen, M.F.A.M., 2004. The effects of edge-line on speed and lateral position: a meta-analysis. Accid. Anal. Prev. 36, 671–682.

Vaniotou, M., 1990. The perception of bend configuration. Res. Transport Securite17, 39–48.

Watts, G.R., Quimby, A.R., 1980. Aspects of Layout that Affect Driver’ Perception andRisk Taking (Rep. 920). Transport and Road Research Laboratory, Crowthorne,UK.

Weir, D.H., McRuer, D.T., 1968. A theory for driver steering control of motor vehicles.Highway Res. Rec. 247, 7–28.

Wegman, F., 2007. Road traffic in the Netherlands: relatively safe but not safeenough! Improving traffic safety culture in the United States: the jour-ney forward, April 2007, AAA Foundation for Traffic Safety. http://www.aaafoundation.org/pdf/Wegman.pdf.

The impact of perceptual treatments on lateral control: A study using fixed-base and motion-base...

Documents

Transcript of The impact of perceptual treatments on lateral control: A study using fixed-base and motion-base...

Base XML - cloudfront.net

9198-BASE NOTICE BROY:5121 BASE NOTICE BROYEUR

Consensus Auditory-Perceptual Evaluation of Voice

Automatic synthesis of high-speed processor simulators

Tugas d Base

Sistema base

Fossil Fuel Power Plant Simulators for Operator Training

Documento di Base

Flight Simulators. A Review of the Research and Development

Base paper

The state of peer-to-peer network simulators - CiteSeerX

BASE DE SUCURSALES

Embodiment and Perceptual Crossing in 2D

Comparing perceptual learning across tasks: A review

Testing Multi-GNSS Equipment: Systems, Simulators, and the Production Pyramid

Distributors of Flight Simulators - Asia Aviation Services

VIABILITY OF USING ENGINE ROOM SIMULATORS FOR ...

Perceptual Surgical Knife with Wavelet Denoising - MDPI

On the perceptual organization of speech

Digital Video Image Quality and Perceptual Coding