Driver behaviour on winter roads : A driving simulator study

(1)

Driver Behaviour on

Winter Roads

A Driving Simulator Study

Carl-Gustaf Wallman (e lea ) ou. <C ) , q aand le & as-as CG Stamp

Swedish National Road and I Transport Research Institute

(2)

VTI rapport 419A - 1997

Driver Behaviour on

Winter Roads

A Driving Simulator Study

Carl-Gustaf Wallman

Swedish National Roadand

'Transport Research Institute

(3)

(4)

Publisher Publication

VTI rapport 419A

Published Project code

_ _ 1997 30154

Swedish National Road and

Project

'Transpart BBSBEI CII I StitUtE Road Surface Status - Effects on Road Users

Author Sponsor

Carl-GustafWallman Swedish Transport and Communications Research Board

Trtle

Driver Behaviour on Winter Roads: A Driving Simulator Study, Part 1

Abstract (background, aims, methods, result)

To optimise road maintenance and operation, road administrators are increasingly using management systems. An important requirement of such systems is the assessement of how different road conditions and measures of maintenance and operation effect road users.

Knowledge of the relationship between road conditions and driver behaviour is often insuf cient. Major factors in uencing a driver s operation ofa vehicle include visual and kinesthetic information about the friction conditions of the road surface. Low friction is an especially obvious problem under winter conditions.

This project attempted to

1 answer the question of whether the simulator environment is suf ciently realistic for experiments with varying road conditions,

and

2 clarify the importance of visual and kinesthetic information for the driver.

The experiment was designed around six scenarios: a road in summer condition, and ve in winter condi-tions with different states of friction.

The results demonstrated that driver behaviour in the simulator was realistic. Regarding speed and lateral position the behaviour was very consistent; there were always signi cant differences in speed levels between the summer scenario and all of the winter scenarios, and no signi cant differences between the winter sce-narios. There were no signi cant differences in the lateral positions. The conclusions are that visual informa-tion is by far the most important for the choice of speed, and that drivers are very poor at evaluating different

friction conditions.

ISSN Language No. of pages

(5)

(6)

Preface

The project presented in this report was part of a three-year research effort, Effects of Road Maintenance and Operation , undertaken at the Swedish National Road and Transport Research Institute (VTI) and nanced by the Swedish Transport and Communications Research Board (KFB).

I am deeply indebted to my colleagues at the VTI for all their encouragement and valuable comments. In particu-lar, I wish to thank Staffan Nordmark, Hakan Jansson, and Mats Lidstrom of the driving simulator team, who very willingly full lled my wishes concerning the experiment s design; Beatrice Soderstrom, for skillfully conducting the simulation runs; Lena Nilsson, for making many relevant and important comments as the of cial reader at the pub-lication seminar; Hans Velin for producing a couple of important graphs; Annette Karlsson, for editing the manu-script; and above all Hakan Alm, who untiringly tried to lend me some of his immense knowledge of experimental design and analysis.

Linkoping, September 1997 Carl-Gustaf Wallman

(7)

(8)

Summary

To optimise road maintenance and operation road ad-ministrators are increasingly using management sys-tems. Assessing the effects of different road conditions and measures of management and operation on road users is an important requirement of such systems. Un-fortunately, knowledge of the relationship between road conditions and driver behaviour is insuf cient.

A major factor effecting operation of a vehicle is the friction force between the tyres and the road surface. Low friction is an especially obvious problem in winter driving conditions. A driver is usually unable to make a direct and correct estimate of the friction level, conse quently indirect information, e. g. visual, kinesthetic, and auditory, is employed while driving.

This project attempted to clarify the importance of visual and kinesthetic information for the driver. Given the dif culties of controlling experiments under real traf-c traf-conditions, it was traf-carried out as an experiment in the VTI driving simulator. The original experiment was de-signed as four scenarios on a 20 km test road:

A. A dry summer road with friction coef cient f = 0.8.

B. A winter road with summer friction, f = 0.8. C. A winter road with mostly summer friction,

f = 0.8, but also with fairly slippery sections, f = 0.25.

D. A winter road with fairly good winter friction, f = 0.4, but with some fairly slippery sections, f = 0.25.

After a preliminary analysis, the experiment was extended with two more scenarios:

E. A winter road with fairly good winter friction, f = 0.4, on the entire road.

F. A winter road with fairly slippery conditions, f= 0.25, on the entire road.

No very slippery conditions (f < 0.25) were chosen. The reasons for this were the lack of previous

experi-ence with studies like this one and concerns that too many of the drivers would not adapt suf ciently to the winter conditions and drive off the road if the friction were very low. The latter would undermine the value of the experiment.

The experiment was performed as a within-subj ects design, i. e. scenarios A, B, C, and D were run by each of twelve subjects, and scenarios A, E, and F were run by each of six additional subjects.

The measured variables were speed, distance, lateral position, yaw angle, steering wheel angle, steering wheel angle velocity, and lateral acceleration.

Driver behaviour regarding speed and lateral position was very consistent over the entire length of the road. There were always signi cant differences in speed lev-els between the summer scenario and all of the winter scenarios, and no signi cant differences between win-ter scenarios. There were nosigni cant differences for the lateral positions.

When behaviour on particular sections of the road was studied, speed relations remained the same, but the lateral position changed with speed and friction values. The other driver variables (yaw, steering wheel motion, and lateral acceleration) correlated with speed and fric-tion as anticipated, i. e. higher speed and lower fricfric-tion yield larger yaw angles and more steering wheel action. The conclusion is that visual information is by far the most important cue, at least for coef cients of fric-tion greater than 0.25. This probably re ects the fact that vehicle handling in normal situations is not much affected even by a relatively low friction such as 0.25. A natural extension of this experiment would be to do trials with lower friction values.

Some validation was done on an aggregate level, comparing average speed under different road condi-tions, and the simulation results matched measurements taken in real traf c. There is also other evidence that drivers are very poor at evaluating different road condi-tions, resulting in insuf cient adaptation to varying fric-tion values.

(11)

(12)

1 Background

Road networks must be kept in good condition to meet drivers needs for safety and traf cability. Pavement should be even and free of ruts, cracks and other dam-age. To maintain good friction in winter, measures against snow and ice must be undertaken. Currently the limited resources available to road administrators means that the goals cannot always beachieved and that pri-orities must be established: what to do, when, where, and how? To optimise maintenance activities (or at least make good choices) administrators use management systems. These systems require assessing the effects of road conditions and maintenance efforts on road users. Unfortunately, the relationship between road con-ditions and, for example, driver behaviour is not well known. Improved knowledge in this area would certainly lead to better management of roads through more cost-effective measures to improve safety.

Driving a car is a complex task placing high demands on drivers perceptual and cognitive processes. Relation ships between environmental variables are entangled, and when a variable cannot be perceived directly it can only be assessed through cues associated with it in some probabilistic way. Social Judgement Theory (e.g., Hammond et al., 1986) may be used to describe this kind of relationship between drivers and the road environ ment.

Driver behaviour on winter roads is in contrast to summer conditions affected by darkness, frost and

Ecological side Validity

snow. The main factor in uencing a driver s operation of a vehicle is the friction (or lack of it) between the tyres and the road surface. It is unlikely that a direct estimate of the friction can be made; instead the driver uses different cues to obtain an indirect, expected value. Figure 1.] presents a so called lens model of the rela-tionship between friction, cues and driver judgement.

Three cues affecting different senses are shown in the gure: a visual cue (white road surface), a kinesthetic cue (the vehicle skids), and an auditory cue (the sound from the tyres is not normal). However, ecologically the relation between each cue and the friction is not per-fect, so the probability is less than one. Subjectively drivers are inconsistent or do not fully utilise the cues, so the probability here is also less than one. The com-pound effect of validity and utilisation produces the achievement: the extent to which the driver makes a cor-rect judgement.

Of course, many other cues more or less subtle -are involved: weather, weather reports, temperature in-formation, the road conditions the driver perceived while walking to the parking lot, etc. Complete knowledge of the cognitive process even for such a comparatively simple task as estimating friction seems almost impos-sible to achieve. However, through confined and con-trolled studies valuable knowledge about driver behav iour in relation to different cues of low friction on win-ter roads should be attainable.

Subjective side Utilisation Low Friction Skidding > Driver

A

Sloshing Sound Achievement Figure 1.1

VTI RAPPORT 419A

Lens Modelfor a Driver '5 Experience ofLow Friction.

(13)

2 Problem

Better knowledge ofhow drivers adapt to and deal with impaired conditions on winter roads is needed. This knowledge will assist road authorities in undertaking the correct measures at the best time.

To measure driver behaviour it is necessary to de-termine which variables and cues to study, and to what extent the experiment can be controlled. The latter is fundamental.

An experimental study conducted under real traf c conditions would be almost completely uncontrolled. Weather, precipitation, and the presence of other traf c could not be controlled. Road surface conditions could be controlled to some extent through operational meas-ures, but the friction would still vary along the road and there would be no way to track it at all places all the time. Furthermore, comprehensive experimental equip-ment comprising sensors, computers, etc., would be needed in the car. Even ifthese problems could be solved it would be necessary to maintain a research group which could be activated on very short notice when the right winter weather conditions occurred.

Safety would be an additional concern. Road admin-istrators must take measures against ice and snow, so the most interesting (and dangerous) road conditions are comparatively infrequent. Exposing drivers and equip-ment to serious hazards is not acceptable in any case, consequently it is desirable to nd an alternative to ex-perimenting under real conditions.

The VTI driving simulator might constitute such an alternative (Nordmark, 1994). In the simulator the ex-perimental situation is as controlled as possible: the road

12

alignment and condition (including friction), other traf-c, and the time of the experiment can be freely cho-sen. Even the most dangerous conditions can safely be simulated. There are problems: the simulated scenery of the road environment is obviously somewhat arti cial and stereotypic; the vehicle movements are simpli ed, for example the acceleration levels are scaled down to 50 percent of real values; there is a delay of40 millisec-onds between the movement of the steering-wheel and the displaying of the picture. The scaling of accelera tion values is probably of limited importance, but the delay in the visual system may result in an unrealistic feeling and also cause nausea simulator sickness for some subjects.

A number of studies have been performed with the VTI driving simulator. Those most closely related to winter driving behaviour concern cognitive load and driving speed (Harms, 1989), driver behaviour in simu-lated fog (Harms, 1991; Alm & Nilsson, 1996), mobile phones (Alm & Nilsson, 1994), incident warning sys-tems (Alm & Nilsson, 1997), and a validation study (Alm, 1995). In all these studies subjects were driving in a seemingly realistic way, with no apparent playing with or testing of the simulation system.

To this point no simulator experiment has addressed the impact of road surface conditions on driver behav-iour. The effect of winter road conditions on speed lev-els has been measured in several eld studies (see sec-tion 7), but only on an aggregate level without detailed consideration of driver behaviour.

(14)

3 Hypothesis

This work is primarily a methodological study intended to answer the crucial question of whether the simulator environment is suf ciently realistic for experiments with varying road conditions.

The hypothesis is that driver behaviour is in uenced by visual and kinesthetic cues in the simulated road en-vironment and that this in uence can be observed through variables related to driver behaviour. If this hypothesis proves correct, the next question would be whether driver behaviour in the simulator corresponds to driver behaviour on real roads.

Variability in driver behaviour throughout the popu-lation is very large, and the same driver does not always

behave in the same manner. Road conditions also vary, even under similar weather conditions. Therefore, it will be dif cult to establish statistically signi cant correla-tions between real and simulated driving. However, com-parative studies across real and simulated environments may be valid.

Driver behaviour will be measured in terms of speed, lateral position, steering-wheel action, yaw movements of the vehicle, and lateral acceleration.

Visual information is obviously the most important for driving.1 It is therefore reasonable to assume that the visual cue will explain the largest part of the vari-ance in driver behaviour.

1 Rockwell (1972) and others state that the driving task is based on visual information to the extent of90%. However, there is no empirical support for this gure (Sivak, 1996).

(15)

4 Definitions and Measures

The following variables were recorded during the simulations: 14 Variable -driving speed distance -lateral position -yaw angle -steering-wheel angle

steering wheel angle velocity

-lateral acceleration

Unit

kilometres per hour (km/h)

metres (m) In

radians (rad) rad

radians per second (rad/s)

metres per second squared (m/sz)

Points on the road were designated by the distance in metres from the starting point. Lateral position was measured from the centre-line of the road to the mid-point of the driver s head. Yaw angle was measured from the centre lines of the road and of the vehicle. Steering-wheel and yaw angles are measured from straight-forward position; positive direction is counter-clockwise.

The sampling frequency was 50 Hertz (Hz) for all variables.

The angle values were converted to degrees at the evaluation.

(16)

5 Method

5.1 The Road

The characteristics of the road used in the experiment were taken from a real rural road, no. 621, south-west of Linkoping. Its length is eight kilometres and width is seven metres. The posted speed limits on two different sections are 70 and 90 km/h. There are 14 bends with radii less than 600 metres.

This road has been used in earlier simulator studies (Harms, 1991; Alm, 1995).

A suitable length for the simulated drive was assumed to be 20 kilometres, with a driving time of about 15 minutes. Consequently, the road was lengthened to 10 kilometres by adding an arti cial section that included a couple of curves with large radii. The road was then doubled so that the geometry of section S and section S+10,000 metres were identical. The posted speed was set at 90 km/h for the whole road.

5.2 The Simulator

The VTI driving simulator is an advanced simulator that includes moving base, wide angle (120 degrees) visual, vibration generating, sound, and temperature regulating systems (Nordmark, 1990; Nilsson, 1993). These ve subsystems can be controlled to give the impression that the driver is in a real car.

The car body used in the simulator was a SAAB 9000. The driving characteristics of the car in this ex-periment were like that ofa vehicle with rear-wheel drive, however, to facilitate the early onset of skidding condi-tions.

5.3 The Subjects

Given the many variables already present in the experi-ment, the subjects were chosen from a homogenous group: males, 25 - 40 years of age, possessing a driv-er s license for at least ve years, and driving at least 10,000 kilometres per year. They also had experience from earlier simulator studies.

Twelve subjects were originally selected to partici-pate in the experiment. A preliminary analysis after four

simulation runs revealed that an extension of the experi-ment would be of interest. Six more subjects were then chosen for a slightly modi ed experiment. The mean age for the 18 subjects was 32.4 i 3.5 years.

5.4 Design and Realisation

Although there are several more or less relevant cues relating to perception of road friction, it was desirable to limit the number present in the experiment. To begin this process the cues were restricted to those perceived from within the car. As the driving task is mainly based on visual information, a white road in winter should be a very strong cue. Kinesthetic information is certainly also important, especially during acceleration or decel-eration and in negotiating curves. Auditory cues like sloshing mud and skidding tyres give valuable informa-tion as well.

These three cues would have been suitable for this experiment. However, the sound system could repro-duce only engine and normal road sounds. There was no possibility of simulating sounds related to surface conditions in this study, such as the sloshing of mud or the skidding of tyres, hence the experiment was limited to visual and kinesthetic cues.

The question was raised as to whether a scenario with no visual cue of slippery conditions should be devised. The presence of black ice could have been simulated, but the risk that many subjects would drive off the road in the rst icybend, rendering the experiment meaning-less, was considered too large.

Two road environments were created. One was a dry black road, set in summer conditions, and one was a very illusory white road with four greyish black wheel-tracks, set in a winter landscape (see Figure 5.1). The tracks were about 0.70 metres wide, the distance be tween their centres was 1.80 metres, and the outer edge of the left track ran 0.73 metres from the centre line of the road.The effect on the track location of drivers tak-ing short cuts in curves was disregarded.

(17)

Figure 5.1

16

The two road environments.

(18)

5.5 Scenarios

Four initial scenarios were established:

A. A dry summer road with friction coefficient f = 0.8.

B. A winter road with summer friction, f = 0.8. C. A winter road with mostly summer friction,

f = 0.8, but also with fairly slippery sections, f = 0.25.

D. A winter road with fairly good winter friction, f = 0.4, but with some fairly slippery sections, f = 0.25.

The friction value of 0.4 corresponds to slush or loose snow but where the tyres have some contact with the pavement. The value 0.25 represents dry, hard-packed snow. The values were chosen after driving tri-als in the simulator.

Scenario A was the reference scenario, with perfect driving circumstances. Scenario B tested the signi -cance of the visual cue. Scenario C introduced slippery sections without prior kinesthetic indication of low fric-tion. Scenario D combined slippery sectionswith nor-mal winter friction which, although good in compari-son to the slippery sections, should provide some warn-ing that lower friction could occur. The sections with low friction were situated where interesting driver be-haviour was anticipated, and always included curves.

After a preliminary analysis two more scenarios were included:

E. A winter road with fairly good winter friction, f = 0.4, on the entire road.

F. A winter road with fairly slippery conditions, f = 0.25, on the entire road.

The preliminary analysis indicated no variation in speeds across scenarios B, C, and D, consequently it seemed interesting to examine the effect of homogenous low friction over the entire driving distance.

No very slippery conditions (f< 0.25) were chosen. The reasons for this were the lack of previous experi-ence with studies like this one and concerns that too many of the drivers would not adapt suf ciently to the winter conditions and drive off the road if the friction were very low. The latter would undermine the value of the experiment. The intention here was to give the driv-ers a fair chance of making successful runs. However, if driver behaviour looks realistic subjectively and speed levels are consistent with field measurements, future research should include experiments with lower friction values.

There was no traf c moving in the same direction as the test driver. Eleven cars were encountered on dif-ferent parts of the road, mostly at the curves. Eight meeting points were randomly distributed across the different curves, and three were subjectively situated on almost straight sections. The meeting points were at 2,100, 4,700, 5,150, 5,800, 6,800, 8,700, 10,950, 12,480, 14,900, 15,150, and 17,900 metres.

The road curvature, low friction sections, and meet-ing points are shown in Figure 5.2. Right-hand and left-hand curves are indicated upwards and downwards from the base-line, respectively. Low friction sections of scenarios C and D are indicated by dotted and continous lines, respectively. Meetingpoints are indicated by M25. The radii given are the smallest in each curve.

(19)

160 820 0 500 1000 1500 /2000 m l I I I I I I I I I I I I I J I I I I I \ / \ / 440 560 LQ _______________1 12 260 2000 M \ 2500 8000 3500 4000 m l _u. I I I I I I I I I I I J I I I I I I \ / 600 270 Lq __________ -_J D I ID I 260 1,00 470 4500 M /-1'£0 \5000 M A 5500 M 6000 m I I I I I I : I I I I : I I I I : J J 1160 V 280 LC.__________________________________________________________________ -_1 ID 1 270 500 6000 6500 M 7000 7500 8000 m I I I I I I I I . I I I I I I I I I I I F] '

880 V

270 _{LC:________________ _.._1} LD 1 8000 8500 M 9000 9500 10000 m I I I I L I I = I I I I I I I I I I I J 160 820 10000 10500 M11000 11500 oo m l I I I I l I I I I : I l I I I I I J I I I \ / \ /

440

560

LC.______ __ ID I 260 12000 \ M 18000 18500 14000 m | I I I I = I I I I I I I I I I I I I I I I

\_/

\

600

/

_C___________-37. _.

ID I 260 1400 1 70 111500 / 14305000 A 15500 16000 m m 1 1 I 1 1 1 1 1 1 1 1 1 1 1 1 1

To

LC. _____ _..._l 270 500 16000 16500 17000 17500 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 180100 m \____J ' 880 V 270 LC:__________ ___1 LQ__I 18000 18500 19000 19500 20000 712 I I I l I I I I I I I I I I I I I I I I I

Figure 5.2 Road Curvature, Low Frictian Sections, and Meeting Points.

(20)

Table 5.1 Co-ordinatesfor Fairly Slippery Sections, f = 0.25. Scenario C Scenario D 800 - 1,150 1,850 2,310 2,300 - 2,600 3,300 - 3,900 Section (m) 4,550 6,000 4,550 - 6,000 6,970 - 7,400 6,700 7,400 11,800 12,600 10,300 11,690 14,800 - 15,000 12,300 - 12,600 16,400 16,700 16,560 - 16,700

Table 5.1 presents the length co-ordinates for the low friction sections.

As can be seen, the sharpest bends are in the sec-tions around 500 and 10,500 metres. As the rst curve occurs at the beginning of the simulation only the sec-ond is suitable for behavioural studies. In scenario D, this curve is in a section with low friction. Interesting driving behaviour could be anticipated, especially as the small radius is rather unexpected after a long straight stretch.

5.6 Simulation Runs

The experiment was designed as a within-subjects study. The rst twelve subjects drove scenarios A, B, C, and D (Design I), and the latter six drove scenarios A, E, and F (Design II). Both alternatives were

counterbal-anced to control for practice effects. The plans for counterbalancing are shown in Tables 5.2 and 5.3. Prior to the simulation runsevery subject drove a two kilometre practice road, half in summer conditions and half in winter, to become accustomed to the simulation environment.

The instructions to the subjects were quite simple: You are to drive a twenty kilometre road under sum-mer and winter conditions. Drive as you usually do under those circumstances . The subjects were also told that in spite of the SAAB body, the car had rear-wheel drive. Every run was video-recorded.

In the rst part of the study subjects drove two sce-narios, then had a coffee break. In the second part sub-jects had their coffee after the three runs. No cases of

nausea were reported. Table 5.2 Plan for Counterbalancing Design 1.

Driver no. Run 1 Run 2 Run 3 Run 4

1, 5, 9 B A D C

2, 6, 10 C D A B

3, 7, 11 D B C A

4, 8, 12 A C B D

Table 5.3 Plan for Counterbalancing Design 11. Run 1 Driver no.

1

(21)

After the procedure the subjects were paid 500 SEK. Two mistakes occurred in Design I: the lateral ac-Only one subject skidded offthe road. That happened celeration was not measured for the rst six subjects,

in the sharp bend in scenario D_ The other subjects ne and the yaw angle values WCI C I BCOI dCd tOO few

gotiated that curve successfully, with more or less ef deCimaIS, resulting in mundng errOI

S-fort.

(22)

6 Results

The results were analysed on two levels: the aggregate to determine if the cues had any impact on mean speed and mean lateral position over the entire 20 kilometre test road, and in detail for some sections of the road to de-termine if and how behaviour is in uenced by friction level, curves, and oncoming traf c.

Subject no. 8 was excluded from the general analy-sis because he drove off the road in scenario D. To maintain the counterbalance drivers 9, 10 and 11 were also excluded, leaving only eight subjects. The driving pattern for subject no. 8 did not indicate that his behav-iour was noticeably effected after he drove offthe road, so for the detailed studies all subjects were included.

A preliminary analysis was done after the rst four subjects completed their drives. The difference between the mean speeds in scenarios B, C, and D were so small that two additional scenarios, E and F, were devised with the friction coef cient for the whole test road set at 0.4 and 0.25, respectively.

The sampling frequency was 50 Hz during the simu-lation. For the rst part of the analysis all data concern-ing speed and lateral position were used. The mean val-ues and standard deviations were generated from the simulation together with plots of driver behaviour. Stand-ard statistical programs for one-way ANOVA and Tukey

HSD tests were used. The Tukey tests were undertaken to control familywise errors, occurring when all possi-ble pairwise comparisons are performed simultaneously. For the detailed studies the sampling frequency was reduced to one sample for every ten metres to obtain a manageable amount of data. The reduction produced a difference of less than 0.4 percent in mean speed; for swift courses involving steering-wheel movements while skidding, the relative difference might be greater. How ever, this variance was not considered too annoying for making comparisons.

Special routines were coded in Q Basic for data processing and ANOVA tests for a within-subjects de-sign. The same Tukey HSD test as for the rst part was used here. The level of signi cance was 0.05.

To improve the readability of this report, few data concerning the statistical tests are provided in this sec-tion. The variance analyses and other statistics are as-sembled in an Appendix.

6.1 Aggregate Analysis

A typical driving pattern can be seen in Figure 6.1, where speed and lateral position are plotted along the whole route (in the driving simulator a lateral position to the right of the centre line has a negative sign).

(23)

Subject no: 6 Sequence no: D Speed(Km/h) 120[ Mean: 79.1 S: 7.1 110 l I 0 500 1000 1500 2000 2500 3000 3500 4000 Distanoe(m) 3 F Mean: -1.49 S: .37 CD Distance (m) _3_.._ Speed(Km l) 120' 110' Mean: 88.7 S: 6.5 5000 5500 6000 6500 7000 7500 8000 8500 9000 9500 10600 Distance(m) Side pos(m) 3 -r- _{Mean: -1.49 S:} _.24 l l l I l c J' g g I i I I l I 5000 5500 6000 6500 7000 7500 8000 8500 9000 9500 10600 1 _._- Distance(m) _2 d z/ W W

Figure 6.1 An Example ofDriving Pattern: Speed and Lateral Position.

(24)

Subject no: 6 Sequence no: D Speed(Km/h) 120' Mean: 79.1 S: 5.9 1 10* l I I I I 10000 10000 11000 11500 12000 12500 13000 "13500 ~ 14000 ~ 145'00 15000 Distance(m) Side pos(m) 3" Mean: -1.41 S: .39 .m . . . 0 1 I : . 101100

11000 11500 12500 ' 15000 Speed(Km/h) 120 Mean: 84.8 S: 7.7 110' 15000 15500 18000 16500 17000 17500 18000 18500 19000 19500 20000 Distanoe(m) Side pos(m) 3-5 Mean: 4.50 S: .31 17500 18000 18500 19000 19000 20000 Distanoe(m)

Total mean speed: 82.8 S: 7.9 Total side pos.: -1.47 S: .34

(25)

The average speed for each subjeet in each scenario The kinesthetic differences are fully noticeable when is shown in Figure 6.2. There is remarkable consistency driving under different friction conditions, but they seem in the behaviour ofthe different drivers. The differences t0 be Of practically no importance when the results are

between the Winter scenarios are relatively small, indi- Viewed OVcr the WhOlc rOUtc, chr1 for a friction coef -cating that the visual cue is by far the most important. Cient as low as 0.25.

Mean speed Mean speed

(Km/h) (Km/h) A 110 " 110 100 .90 80 r 70 I I I 70 , I A B C D A E F Scenario ' Scenario

Figure 6.2 Average Speedfor Each Subject and Each Scenario: Design I and II. (A dotted line is usedfor the driver who drove ofthe road).

(26)

Table 6.1a Mean Speed and Mean Lateral Position: Design I, 8 Subjects.

Scenario A B C D

Speed (km/h) 95.9 $ 2.8 84.8 $ 5.0 84.3 $ 4.0 83.5 $ 3.7 Lateral -1.39$0.13 1.32$0.10 1.31 $0.13 -1.31 $0.12 position (m)

Table 6.1b Mean Speed and Mean Lateral Position: Design 11, 6 Subjects.

Scenario A E F

Speed (km/h) 98.2 $ 8.9 82.0 $ 7.6 81.2 $ 5.6 Lateral 1.35$0.14 -1.31 $0.11 -1.31 $0.09 position (m)

The mean speeds, mean lateral positions, and stand-ard deviations for each scenario are indicated in Tables 6.1a and 6. lb. In both Design I and Design II the speeds in scenario A differ signi cantly from the speeds in the other scenarios. There are no signi cant differences between the winter scenarios B, C, and D in Design I and between E and F in Design II. There are no signi -cant differences at all between the lateral positions. Yet, there is a tendency to keep closer to the centre of the road during winter conditions. This might be an effect of the visible tracks on the winter roads.

6.2 Detailed Analysis

In the detailed analysis driver behaviour on a number of short sections of the road was studied. It was assumed that three factors would have an in uence on the driver response variables: rst, the visual impact and friction coef cient ofthe scenario, second, the alignment of the road as a left hand curve, right-hand curve, or straight section, and third, the presence of oncoming vehicles. The variables were distance, speed, lateral position, steering-wheel angle and velocity, yaw angle, and lat-eral acceleration. From these data the following com-putations were generated for every section and subject as a basis for the variance analysis: mean speed, mean lateral position, summarised differential changes of steering-wheel angles, mean steering-wheel angle, vari-ance of steering-wheel angle (calculated per subject), mean steering-wheel angular velocity, summarised dif-ferential changes of yaw angles, mean yaw angle, vari-ance of yaw angle (calculated per subject), and mean lateral acceleration. All data pertaining to steering-wheel and yaw motion were treated as absolute values. The summarised changes thus gave the total steering-wheel motion and the total yaw motion, independent of direc-tion.

There are some points to consider when studying the

VTl RAPPORT 419A

results. The variables arenot independent, but more or less correlate with speed and the prevailing friction. The results are therefore often coherent and obvious, for instance high speed or low friction yields large yaw, while low speed or high friction yields small yaw.

The analysis ofvariance in some ofthe variables may be questionable, especially steering-wheel angle and yaw angle. These variables have no homogenous within-group variances (which will bias the F-test) and should perhaps have been analysed in another manner. Steer-ing-wheel and yaw angle variances may add something to our knowledge of driver behaviour, however, and as they are not the major points of this study, these analy-ses have been retained.

The detailed analysis of driver behaviour is broken down into four studies: short sections, consisting of single curves; longer sections, consisting of reverse curves; straight sections; and the sharp and slippery bend mentioned in section 5.4.1.

Some general conclusions can be drawn prior to a detailed examination of the different sections. There is a common tendency for high speed or low friction to be associated with greater yaw angles and steering-wheel activity, and this seems realistic even if signi cant dif-ferences are not always established.

Decreases in speed in the winter scenarios partly compensates for lower friction. Thus, for Design I, the behaviour at scenario A (where f= 0.8) is rather similar to that at scenario D (where f = 0.4). The behaviour at scenario B (in which the friction is high and speed is low), differs from all the other scenarios, except for those parts of C where f = 0.8. On the other hand, be-haviour where C and D both have f = 0.25 is very simi-lar.

This suggests that the behaviour in scenarios A and E in Design II should be similar. In these cases there are greater speed differences between summer and

(27)

ter conditions so the behaviour at scenarios A, E, and F differ more. However, the distinctions are always con sistent relative to speed and friction.

An interesting point is that the variance of steering-wheel angle is larger the lower the friction. Steering-wheel activity is evidently more pronounced in low fric-tion condifric-tions.The results may be studied more closely in the Appendix.

6.2.1 Short Sections

The features of the short sections are indicated in Table 6.2.

It is common behaviour to take short cuts in curves.

In left-hand curves oncoming vehicles should impede such behaviour, consequently those curves appear most frequently in the study of the short sections.

The analysis of variance disclosed that the mean speeds in both A scenarios are signi cantly higher than in scenarios B, C and D, and scenarios E and F. There are never any differences among B, C and D, or E and F.

For Design 1, low speed and high friction (as in sce-nario B) should imply lesser steering-wheel actions and yaw angles than for either higher speed (as in A) or lower friction (as in C and D). This assumption does hold; there is always a signi cant difference among scenarios for these variables. Most of the difference is between B and one or more of the other scenarios. However, in sections where B and C have the same friction no dif-ference can be discerned between them. For the lateral

position, only sporadic signi cant differences are found. Paired comparisons between section S and S + 10,000 metres for the same scenarios are interesting: if the friction conditions are the same, the consistency of driver behaviour can be tested; if the friction conditions are different, signi cant differences in the behaviour may appear

Driving behaviour seems to be very consistent when friction conditions are constant. Unequal friction con-ditions, on the other hand, produce signi cant differ-ences in steering-wheel and yaw variables. Oncoming vehicles produced no signi cant impact on lateral posi-tion.

Design 11 lacks scenario B, consequently there are few signi cant differences in the steering-wheel vari-ables, but there are still signi cant differences in the mean yaw angles. This is natural given that increased yaw angles are associated with decreasedcoef cients of friction.

Paired comparisons of two different sections are only tests of consistency, for the friction is always constant within each scenario. Like Design I, the driving behav iour is very consistent.

6.2.2 Long Sections

The features of the long sections are indicated in Table 6.3. The curves within each section are separated by straight segments of about 100 metres.

Table 6.2 Data for the Short Sections.

Section 2,340- 12,340 5,060 15 ,060 2,540 12,540 5,240 15,240 Smallest radius (m) L 260 L 260 L 270 L 270

(L=left, R=right)

Scenario with low C C, D C, D friction, f=0.25 Oncoming vehicle M M M Section 6,720- 16,720- 7,050 17,050-6,960 16,960 7,300 17,300 Smallest radius (m) L 265 L 265 R 260 R 260 (L=left, R=right)

Scenario with low D C, D friction, f=0.25

Oncoming vehicle M

(28)

Table 6.3 Datafor the Long Sections. Section 4,550 14,550- 6,000 16,000 6,000 16,000 7,500 17,500 Smallest radii (m) L 440 L 440 L 840 L 840 (L=left, R=right) R 425 R 425 R 485 R 485 L 270 L 270 L 265 L 265 R 250 R250 R 260 R 260 Parts with low C: 4,550- C: 14,800- C: 6,970 C: 16,200-friction, f=0.25 6,000 15,000 7,400 16,700 D: 4,550- D: 6,700- D: 16,560-6,000 7,400 16,700 Oncoming vehicles M: 4,700 M: 14,900 M: 6,800 M: 5,150 M: 5,800 The purpose of this part of the study was to exam-ine driving patterns in a sequence of curves consisting ofreverse curves combined with sections with low fric-tion.

In Design I there are signi cant differences for every variable in every section except for steering-wheel an gular velocity, where no difference can be observed. In section 4,55 0 6,000 metres, the steering-wheel and yaw variables in scenarios C and D are very similar (as they should be) and B differs signi cantly from the other scenarios. In section 14,550 16,000 metres, the result is more complex: B resembles C, A resembles D concerning steeringwheel actions, and B differs signi -cantly from A, C, and D on the yaw variable.

The lateral position is very consistent. There is prac-tically no difference between B, C, and D (about 1.30 metres to the right ofthe centre line), while the position in A is about 1.40 metres.

Paired comparisons between scenarios C and D and sections 4,550 6,000 and 14,550 16,000 allow evalu ation ofthe differences among friction coef cients 0.25, 0.4 and 0.8. There are no differences in speeds and lat-eral positions whatsoever, but for every other variable there are signi cant differences between f = 0.25 and 0.8 and between f= 0.25 and 0.4. Comparisons between f = 0.4 and 0.8 produce signi cant differences only on mean steering-wheel angle and the variances for steer-ing-wheel and yaw angles.

In Design II the signi cant differences are found in the yaw variables, where scenario F differs from A and E. There are no signi cant differences in steering-wheel actions, but the values are lowest for scenario E and highest for scenario A (which are rather close to F). The lateral positions do not differ signi cantly, but the

val-VTI RAPPORT 419A

ues for scenario A are greater than for the other sce-narlos.

Comprehension ofthese results is enhanced by look-ing at an individual drivlook-ing pattern, illustrated in Figures 6.3a and 6.3b. Scenario A is compared to B and D, re-spectively. There is a striking resemblance between A and B, and an apparent difference between A and D, especially in steering-wheel and yaw angles.

6.2.3 Straight Sections

The straight (or almost straight) sections were: 2,600 3,300, 12,600 13,300, 5,500 6,100, and 15,500 16,100 metres.

There are always very small movements of steering-wheel and yaw angle. The only variable of interest is lateral position. For Design I, the positions in scenario A are signi cantly greater than in B, C, and D (except for one case), and the values are close to those for the long sections above. For Design II, there is only one signi cant case, but the distances in A are generally greater than in E and F.

6.2.4 The Sharp Bend

A sharp curve with a radius of less than 180 metres between 10,390 and 10,480 metres appeared rather abruptly after a long, smooth stretch. In scenario D the curve was a bit dif cult to negotiate with a friction co-ef cient of 0.25. In Figure 6.4 the mean speeds are plot-ted through the curve in 40 metre sections from 10,320 to 10,800 metres. The shape of the plots are similar, and the speed and friction levels correlate. Data for Design I is based on 8 subjects because one subject did not succeed in correcting a skid on this curve.

(29)

Speed (Km/h) 120c 100' 6700 Lateral pos (m) 5 -4 .7». 3 -r 1... .2 ~ _3 __ _4 d_. _5 0.. 6750 l l l l I l l I i I I I I I I I I l J 6950 7CKX) 7250 7300 Distance (m) 7200 Lateral acc (mlsz) 5.. 4.... 3... 2-.. 1 0-45» Yaw (deg) 15--107 5... G 6700 5 1 40* 45 7250 7300 Distance (m) 7250 7300 Distance (m) H 7300 72150 I Distance (m) 7000 7050 7100 7150 7200

Figure 6.3a Comparison ofDriving Patterns, Scenario A and B.

(30)

Speed (Km/h) 120* 100 20 Lateral pos (m) 5- 4- 3-2_ 1... 0

If:

.2- 3-_4_ l I 7150 7250 7300 Distance (m) l 7(XX) 7100 7200 p > . ... l I 006750:680068'506900 6950l l L l T l 7050 7100 7150 7200 W 7250_{Distance (m)}i 7300i 7000 . _5_.. Lateral acc (mlsz) .2 a _3_ _4.. 6.. i 7250 7300

7000

7100 "'50 7200

_

Mjstance (m)

St-wheel angle (deg)

45 . _45_ Yaw (de 15 10" 5- 0-67 6 -10 -15 Figure 6.3 VTI RAP 7250 7:300 Distance (m)

g) AI; BL} J %\ 1 ¥ 1 % 1 l 7000 7050 Distance (m)

b Comparison ofDriving Patterns, Scenario A and D.

(31)

110 I I I I I I I I I I I I I 100

90 A

c \

E

B

80 *- C D 70 ' 60 I I 1 I I I l l I I II I 820 360 400 440 480 620 660 600 840 680 720 760 800 Section minus 10000 In 110IIIIIIIIIIIII

100 -A

90

.C \ E X

80 -E

70 F

60 I J I I I I I I I I l I I 320 380 400 440 480 520 580 600 640 680 720 760 800 Section minus 10.000 m

Figure 6.4 Mean Speeds in the Sharp Curve: Design I and Design II. Analysis of the variance in Design I yields

signifi-cant differences for all variables. As always the speed is greatest in scenario A. Otherwise it is scenario D that is different: as it involves a right hand curve, the car skids closer to the centre line than in the other scenarios; there is also stronger and faster steering-wheel action as well as more yaw motion. The driving behaviour is best de-scribed by the mean values ofthe variables as indicated in Table 6.4.

For Design 11, on the other hand, there are signi -cant differences for only lateral position and mean yaw angle.

A paired comparison of scenarios C and D reveals signi cant differences for every variable except steer-ing-wheel angular velocity. A comparison between E and F reveals a signi cant difference only for mean yaw angle.

Individual behaviour in the curve is illustrated in Fig-ure 6.5, where scenario A is compared to D for subject no. 9. The skid was noticed through shifts in lateral position and yaw angle. To correct the skid in D, the subject turned the steering-wheel violently. A closer view of the steering-wheel motion is shown in Figure 6.6. Table 6.4 Mean Values ofDriving Variables in the Sharp Curve.

Scenario A B C D Speed (km/h) 86.6 76.1 74.8 71.3 Lateral - l .66 -1.65 1.64 1.45 position (In) Summarised 131.6 117.0 115.7 243.0 steering-wheel angle (degrees) Mean steering-wheel 24.59 21.29 21.00 24.50 angle (degrees) Variance in steering- 91.5 49.4 53.7 205.6 wheel angle Steering-wheel 9.66 7.85 7.80 15.25 angular velocity (degrees/second) Summarised 9.17 6.45 6.30 17.12 yaw angle (degrees)

Mean yaw angle 0.44 0.19 0.18 1.14 (degrees)

Variance in yaw 0.24 0.11 0.10 1.26 angle

30

(32)

Speed (Km/h)

120 ~

m..-l l I l l l l l I l I

$8300 10850 10400 10350 10500 1(550 10500 10550 10700 10750 10800 10850 10800

Lateral pos (m) Distance (m)

5 . 4 1. 3 -2 .. 1--1 A 1 1 1 1 1 1 1 0 1 i 1 1 1 I I I 1 I 1 i 1p§ipo 10350 1 10450 1 550 10600 2550 10700 10750 10800_d1085n/19900 2 u_ M Distance (111) 6.1.. 44

.51

L Lateral acc (mlsz) 5... 4-.. 3 ... 2 1..

H

A

M

G l l i l l I l l J A ' I 10850 10900 Distance (m) St-wheel angle (deg)

45 1

H

30 . 15--l l J 1 10700 10 \M 10850 10900 Distance (m) ad »

10800

10400 10450 10500

f\

@4

-45J

I

Yaw (deg) 15---15~ 107 5.0.. 0 1 I 1 1 A UL 1 1 1 1 n 1 11 7 71 a 10800 1 1 1 107 10750 10800 10850 10900 _5 __ Distance (m) -10~~ -15 L

Figure 6.5 Comparison ofDriving Patterns in the Sharp Curve, Scenario A and D.

(33)

Yaw (deg)

m.

75-h L

W 10850' 70900

_{Distance (m)}

Figure 6.6 Detailed Plot ofSteering-wheel Action in the Sharp Curve, Scenario A and D.

(34)

7 Other Studies

A complete study ofthe validity ofthe experiment must include the driving pattern on an aggregate as well as a detailed level. To this point detailed validity studies have been impossible, but on the aggregate level comparisons can be made to Swedish, Norwegian, Finnish, and Ger-man eld studies.

In two VTI bulletins Oberg (1994) and Wallman

(1996) have compiled measurements of car speeds un-der different road conditions. For roads 7 metres wide with a typical posted speed of 90 km/h, average speeds are 85 to 95 km/h on dry, bare roads. Speeds are re-duced in winter conditions with ice or hard snow on the road surface, typically by 6 to 10 km/h. There are large deviations from these characteristic values: average speed reductions up to 16 km/h have been recorded, and ice or hard snow is not a very rigorous measure of the road condition.

In a VTI report, Oberg (1978) found that the speed

increase on a newly gritted road was larger than an in-crease corresponding to the improved friction. Drivers seemed to adapt to an expected rather than the real level of friction based on a visual impression.

On road E18 in Norway Ruud (1981) measured mean speeds of 81, 76 and 74 km/h when the friction coef cients were 0.65, 0.40 and 0.25, respectively. On two national roads the speeds were recorded to be 76, 71 and 69 km/h, when the friction coef cients were 0.70, 0.39 and 0.26, respectively. It is appropriate to note that the speed limit was 80 km/h.

Three reports from the Finnish National Road Ad-ministration contain very interesting results. Heinijoki (1994) examined the extent to which drivers take slip-periness into consideration in winter through driver in-terviews and measurements of car speeds. Road slip-periness was measured and divided into four categories:

good grip (f > 0.45), fairly good grip (0.35 < f S 0.45), fairly slippery (0.25 < fS 0.35), and slippery (fS 0.25). The drivers were asked to evaluate the slipperiness on the same scale. Generally, the drivers were poor at evalu-ating the actual road conditions. Less than 30% of the evaluations coincided with the measured values, and more than 27% differed by 2-3 categories. The more slippery the conditions the more evaluations differed from reality, consequently the slipperiness of the road did not have any appreciable effect on driving speed.

Saastamoinen (1993) found that driving speed de clined mostly as a function of wintry weather or reduced speed limits. Road conditions were signi cant, then, only in the case of snowy weather. Compared with good driving conditions, speed decreased by 0 3 km/h when the grip was only fairly good (see above), 3 6 km/h under fairly slippery conditions, and 4 7 km/h under slippery conditions. The speed did not change to any appreciable extent when the conditions changed from fairly slippery to slippery.

Roine (1993) examined driving speed in sharp curves. The average speed was about 6 km/h less in sharp curves under slippery compared with dry or wet road conditions. This result can be compared to the simulated result in the sharp curve: the mean speeds for the slippery scenarios were 13 19 km/h lower than for the summer scenarios.

In Germany Durth, et al. (1989) measured the im-pact of low friction on 7 different roads with a width of 5.2 to 8.5 metres. The average median speed for these roads was 76 km/h on a dry surface, and 45 km/h(!) on a slippery surface. The friction coef cient was not measured, but is presumed to be 0.75 and 0.15, respec-tively.

(35)

8 Conclusions and Discussion

The crucial question formulated in section 2 was Is the simulated environment realistic enough that a driver acts in the same way as under real circumstances? Within the limitations of this project there is strong evidence that the answer is yes. The aggregate speed values do match the results from eld studies described in sec-tion 7, and there is strong agreement on speed reduc-tions during winter condireduc-tions. The reducreduc-tions seem to be closely related to visual impact; the prevailing fric-tion values seem to have a very small effect on the driv ers speed in both simulated and real driving.

The speed levels in the simulator are somewhat higher than for real conditions. This effect has also been no-ticed in other studies, e. g. (Alm, 1995).

In a detailed analysis the different friction levels are re ected in driver variables such that the combined ef-fect of speed and friction give credible changes in lat-eral position, steering-wheel and yaw indicators. Even for subtle modi cations in the simulated driving envi-ronment such as decreasing the friction coef cient from 0.8 to 0.4 or from 0.4 to 0.25, there are consistent and credible shifts in the driving pattern. The subjects also seem to take the experiments very seriously; the inhibi-tions against colliding with oncoming vehicles or driv-ing offthe road seem to be almost as strong as in real ity.

It is very easy to be carried away by this seemingly true behaviour, but one has to bear in mind that the simu-lation experiment is only realistic, not real! Conse-quently, the validity of the experiment has yet to be de-termined on the detailed level. Moreover, while the driv-ing simulator was not intended to simulate adverse con-ditions or reckless driving, it may be interesting to ex-amine its capabilities concerning those matters.

Are there any alternatives to simulation for performing a study like this? In section 2, experiments under real traf c conditions were excluded because of control and safety reasons. Such studies should nevertheless be undertaken in the future to calibrate and validate the results of the simulation. Another possibility might be

34

to use skid training grounds, but the road and traf c environment is lacking. This would be, however, a suit-able way of studying drivers techniques for skidding correction and could also be used for calibration and vali-dation purposes.

Some re ections on the subjects choice of speed levels are appropriate. When driving on dry summer roads the speed level is probably chosen on the basis of legality and comfort rather than safety. The utilised fric-tion is usually much lower than what is available.

In winter conditions speeds decrease because safety considerations become much more relevant. However, it is evident that the prevailing level of friction has little to do with the choice of speed, at least for the friction interval in this study. The adaptation of speed to low friction levels is very poor. This probably re ects the fact that even for a comparably low friction coef cient such as 0.25, the handling of the vehicle is not very aggravated during normal driving. One may be startled by a sudden event like skidding in a sharp curve and mo-mentarily decrease speed, but apparently there are no persisting effects on driving behaviour.

The conclusion: friction has to be so low that nor mal manoeuvrability is restricted to effect the chosen speed level. Subj ectively experienced, this behaviour also seems to be present in real traf c.

A briefexample may provide additional perspective. If the speed on a summer road is selected to maintain a safe braking distance , in the experiment this distance would be about 45 metres. On a winter road this value implies that the average subject adapt to a coef cient of friction of about 0.6 with respect to the decreased speed. Such a high friction value is not very likely for a typical white winter road.

It should be noted that there is more than one way for drivers to deal with poor road conditions: they can also get more attentive. This may partially explain the poor speed adaptation to different friction levels, for instead of decreasing their speed drivers may be increas-ing their attention.

(36)

9 Future Research

This experiment indicates that car drivers adapt their speed very poorly to slippery road surface conditions, at least down to a friction coef cient of 0.25, a result supported by earlier eld studies. A natural continuation of the research would be to perform the same experi-ment with lower coef cients of friction to nd out where the choice of speed really starts to become affected and how it changes in response to successively lower fric-tion.

Additional future studies could include whether a driver s attention increases with decreasing friction levels,

the signi cance of lateral variations in the friction level across the road, the importance of auditory cues, and variation in vehicle characteristics, e.g. ABS brakes.

In this future research the driver sample should be successively widened to obtain a more representative sample of the driver population. Also, through the ac-quisition of a car with proper experimental equipment it is now possible to make validation studies at VTI on the level of detailed driver behaviour.

(37)

10 References

Alm, H: Driving Simulators as Research Tools: A Validation Study Based on the VTI Driving Simu-lator. DRIVE 11 Project V2065 GEM. Linkoping. 1995.

Alm, H, Nilsson, L: What is the Optimal Level of Detail in a Message from an Incident Warning System?: A Comparison Between the PORTICO and the MELYSSA System. Accident Analysis and Prevention, submitted for publication 1997. Linkoping. 1997.

A1m., H, Nilsson, L: Effects of a Vision Enhancement System on Drivers Ability to Drive Safely in Fog. Proceedings of Vision in Vehicles V, Glasgow, 1993, pp 263-271. VTI sartryck No. 264. Linkoping. 1996. Alm., H, Nilsson, L: Changes in Driving Behaviour as a Function of Handsfree Mobile Phones: A Simulator Study. AccidentAnalysis and Prevention, Vol 26, No. 4, pp 441-451. VTI Sartrka No. 221. Linkoping. 1994.

Durth, W, Hanke, H, Levin, C: Wirksamkeit des Stra enwinterdienstes auf die Verkehrs-sicherheit und die Wirtschaftlichkeit des Verkehrsablaufes. Forschung Stralienbau und Stra enverkehrstechnik. Heft 550, 1989:]. Bundesminister fiir Verkehr, Abteilung StraBenbau. Bonn. 1989.

Hammond, K R, Stewart, T R, Brehmer, B, Steinmann, D 0: Social Judgement Theory, in Arkes, H R, Hammond, K R, Judgement and Decision Mak-ing. Cambridge University Press. Cambridge. 1986. Harms, L: Experimental Studies of Variations in Cognitive Load and Driving Speed in Traf c and in Driving Simulation. The III Conference on Vi-sion in Vehicles, Aachen, 1989. VTI sartryck No. 148. Linkoping. 1989.

Harms, L: The In uence of Sight Distance on Sub-jects Lateral Control: A Study of Simulated Fog. The IV Conference on Vision in Vehicles, Leiden,

1991. VTI Sartrka No. 173. Linkoping. 1991. Heinijoki, H: Kelin kokemisen, rengaskunnon ja

rengustyypin vaikutus nopenskayttéiytymiseen. (In uence of the Type and Condition of Tyres

36

and Drivers Perceptions of Road conditions on Driving Speed.) Finnish Road Administration, FinnRA reports 19/1994. Helsinki. 1994.

Keppel, G: Design and Analysis. A Researcher s Handbook. Prentice-Hall. Englewood Cliffs. 1991. Nilsson, L: Behavioural Research in an Advanced Driving Simulator: Experiences of the VTI Sys-tem. Human Factors and Ergonomics Society, 37th Annual Meeting. Seattle. 1993. v Nordmark, S: The VTI Driving Simulator: Trends and

Experiences. Proceedings of Road Safety and Traf-c Environment in Europe. Gothenburg. 1990. Nordmark, S: Driving Simulators, Trends and

Expe-riences. RTS 94 Driving simulation Conference, Paris, 1994. VTI Séirtryck No. 204. Linkoping. 1994. Rockwell, T: Skills, Judgement and Information Acquisition in Driving. Human Factors in Highway Traf c Research, pp 133 164. New York. 1972. Roine, M: Kuljettaja kiiyttiiytyminen kaarreja

jonoajossa. (Driver Behaviour in Sharp Curves and Queues on Main Roads.) Finnish Road Admin-istration, FinnRA reports 87/1993. Helsinki. 1993. Ruud, H.H: Kjorefart pa saltede och usaltede veger:

malinger i Akershus og Vestfold 1980 og 1981. Transportekonomisk Institutt. Oslo. 1981.

Saastamoinen, K: Kelin vaikutus ajokiiyttaytymiseen ja liikenne virran ominaisuuksin. (Effect of Road Conditions on Driving Behaviour and Properties of Traffic Flow.) Finnish Road Administration, FinnRA reports 80/ 1993. Helsinki. 1993.

Sivak, M: The Information that Drivers Use: Is it Indeed 90% Visual? Perception, Vol 25 No. 9, pp

1081 1089. London. 1996.

Wallman, C-G: Effektberakningar till Lathunden . Hastighetsreduktioner och bransleforbrukning vid olika vaglag. VTI notat No. 71. Linkoping. 1996. Gberg, G: Effekter av sandning. VTI rapport No. 164.

Linkoping. 1978.

Gberg, G: Vadrets och viiglagets inverkan pa personbilshastigheten. VTI notat No. 62. Linkoping. 1994.

(38)

Appendix

Page 1 (19)

Statistical Analysis

This appendix presents results from the statistical analyses of the various road sections.

The analysis of variance for the within-subjects design was done according to Keppel ( 1991), chapter 16. Special computer programmes were designed from the

computational formulas to calculate the means, variances, and F-values for the

various driving variables.

The sensitivity of the experiment was estimated by the (1)2 index (see e. g. Keppel (1991), chapter 4). The size of an effect in terms of (1)2 could be described as:

a small effect when (1)2 = 0.01 a medium effect when (02 = 0.06

a large effect when (02 = 0.15.

The (02 - index for within subjects design was estimated by formulas proposed by Keppel.

The experiment has a very high power; several (1)2 values are in the interval 0.5 0.8. This re ects the strength of the visual cue.

The results of the analysis of variance for each variable are shown in tables for each analysed section of the road.

A standard Tukey HSD test was also done for every variable. In the tables significant F values are denoted by an asterisk (*) where the Tukey test indicates no significant difference.

Analysis of variance requires that the within group variances are homogenous or the F-test will be seriously biased. A simple check is to look at the quotient of the maximum and minimum within group variance. If this value is less than 9, the homogeneity is probably sufficient. This is generally the case for the variables in this analysis except for the variances of steering-wheel and yaw angle.

(39)

Appendix Page 2 (19)

Tables over F-values, level of significance 0.05

Within-subjects design

Test Degrees of freedom F denominator nominator A - B - C - D 3 33 2.88 12 subjects A - B - C - D 3 21 3.07 8 subjects Paired comparison 12 subjects 1 11 4.84 Paired comparison 8 subjects 1 7 5.59 A - E - F 2 10 4.10 6 subjects Paired comparison 6 subjects 1 5 6.61

(40)

Appendix

Page 3 (19)

Section: 2340 - 2540, Left-hand curve

A B C D F -value Omega2 Speed mean 91.78 77.70 76.32 77.37 44.93 0.73 variance 18.64 44.26 27.15 26.73 Lateral mean -0.81 0.92 -0.95 -0.95 1.03 0.02 position variance 0.07 0.10 0.07 0.10 Sum of mean 57.15 56.96 60.59 58.63 0.63 -0.02 st wh angle variance 114.37 214.66 154.30 72.54 Mean st- mean 18.23 15.87 18.25 17.58 12.44 0.42 wh angle variance 1.00 1.96 3.61 2.61 Variance mean 49.30 36.83 58.69 44.50 5.86 0.23 st-wh angle variance 198.01 52.58 615.85 140.20 St-wh mean 7.84 6.91 7.06 7.64 0.53 -0.03 angular vel variance 6.51 7.49 9.18 5.42

Sum of mean 4.34 2.39 4.49 4.01 6.49 0.26 yaw angle variance 1.34 1.72 5.06 1.85

Mean mean 0.38 0.13 0.85 0.41 74.39 0.82 yaw angle variance 0.01 0.01 0.05 0.01

Variance mean 0.16 0.06 0.36 0.14 12.13 0.41 of yaw ang variance 0.00 0.00 0.06 0.00

A B C D F-value Omega2 Speed mean 89.33 79.28 76.03 74.36 41.80 0.72 variance 29.07 42.14 35.49 28.92 Lateral mean -0.96 -0.95 1.03 0.99 0.75 0.02 position variance 0.04 0.07 0.05 0.06 Sum of mean 65.65 55.39 65.79 64.08 2.36 0.08 st-wh angle variance 173.42 169.89 414.46 153.74 Mean st- mean 17.68 16.09 17.81 17.62 9.47 0.35 wh angle variance 1.86 1.56 2.52 3.04 Variance mean 61.67 38.35 57.07 46.61 7.04 0.27 st-wh angle variance 253.72 51.73 793.03 251.37 St wh mean 8.00 6.58 8.90 7.11 2.25 0.07 angular vel variance 4.30 3.73 21.43 8.77

(41)

Appendix

Page 4 (19)

A B C D F-value Omega2 Speed mean 95.26 83.24 81.58 81.26 37.27 0.69 variance 20.54 55.53 26.81 25.93 Lateral mean -1.06 -1.05 -1.11 -1.16 1.70 0.04 position variance 0.03 0.08 0.03 0.10 Sum of mean 59.64 52.47 78.45 68.90 1.96 0.06 st-wh angle variance 299.30 128.12 2978.47 604.28 Mean st- mean 15.78 14.19 16.60 16.27 18.46 0.52 wh angle variance 0.90 1.26 2.38 1.58 Variance mean 82.30 56.13 113.24 103.11 4.75 0.19 st-wh angle variance 447.64 126.98 6522.60 1879.27 St-wh mean 8.42 6.96 9.88 8.47 1.33 0.02 angular vel variance 10.52 9.35 30.15 12.57

A B C D F -value Omega2 Speed mean 91.76 84.99 80.79 80.34 34.16 0.67 variance 13.63 46.73 30.10 39.46 Lateral mean -0.98 -1.01 0.97 1.08 1.45 0.03 position variance 0.02 0.08 0.08 0.05 Sum of mean 51.28 49.99 50.42 52.95 0.29 -0.05 st-wh angle variance 75.36 95.03 45.72 237.94 Mean st- mean 15.06 14.31 13.56 15.41 17.66 0.51 wh angle variance 0.72 1.57 0.53 2.19 Variance mean 73.58 60.78 54.08 65.41 10.02 0.36 st wh angle variance 183.83 108.29 83.18 328.33 St wh mean 7.52 6.61 6.68 7.50 0.43 -0.04 angular vel variance 2.50 6.73 3.31 18.38

Mean mean 0.32 0.19 0.22 0.28 4.56 * 0.18 jaw angle variance 0.00 0.01 0.03 0.02

(42)

Appendix Page 5 (19)

A B C D F-value Omega2 Speed mean 91.03 80.97 80.67 78.89 23.29 0.58 variance 27.08 56.35 39.89 35.50 Lateral mean 1.09 -0.94 -0.87 -1.02 4.05 * 0.16 position variance 0.02 0.10 0.13 0.12 Sum of mean 66.42 57.63 62.83 90.00 6.94 0.27 st-wh angle variance 250.59 150.02 140.16 1492.50 Mean st- mean 19.17 17.06 16.79 20.00 31.61 0.66 wh angle variance 1.53 2.10 1.38 3.84 Variance mean 40.36 33.94 36.90 57.61 16.66 0.49 st-wh angle variance 112.04 54.10 15.58 315.50 St-wh mean 7.44 5.39 7.14 8.19 2.26 0.07 angular vel variance 3.72 3.67 6.18 23.94

A B C D F-Value Omega2 Speed mean 90.05 79.90 78.76 76.48 55.73 0.77 variance 25.20 58.56 30.93 45.58 Lateral mean 1.00 -0.91 0.97 -0.87 0.63 -0.02 position variance 0.15 0.09 0.11 0.10 Sum of mean 57.96 58.82 61.97 60.40 0.49 -0.03 st wh angle variance 89.28 175.62 270.19 122.80 Mean st- mean 19.06 16.79 16.50 17.97 29.57 0.64 wh angle variance 1.30 2.11 1.02 3.16 Variance mean 34.02 34.41 34.44 38.85 1.50 0.03 st-wh angle variance 63.64 23.80 25.08 122.92 St-wh mean 6.00 6.38 6.26 5.90 0.17 0.05 angular vel variance 2.91 6.98 7.49 2.34

Variance mean 0.18 0.12 0.14 0.18 3.07 * 0.11 of yaw ang variance 0.01 0.01 0.02 0.01

(43)

Appendix

Page 6 (19)

Section: 7050 - 7300, Right-hand curve

A B C D F-value OmegaZ Speed mean 91.91 80.65 79.67 79.41 41.57 0.72 variance 22.56 49.97 38.40 20.89 Lateral mean -1.82 -1.60 1.52 1.55 11.28 0.39 position variance 0.07 0.07 0.07 0.07 Sum of mean 63.41 57.87 83.17 94.63 2.08 0.06 st-wh angle variance 165.47 151.43 1280.10 7723.15 Mean st- mean 20.34 18.08 21.48 21.31 26.24 0.61 wh angle variance 1.73 1.96 5.34 3.41 Variance mean 40.27 29.13 50.23 56.26 3.47 * 0.13 st wh angle variance 132.74 50.78 418.51 2152.94 St-wh mean 6.22 6.21 9.06 11.09 1.55 0.03 angular vel variance 3.07 4.86 38.55 198.59

Section: 17050 - 17300, Right-hand curve

A B C D F-value OmegaZ Speed mean 90.53 80.57 79.23 78.13 41.30 0.72 variance 25.69 41.15 31.49 34.17 Lateral mean -1.81 1.63 -1.60 -1.59 8.39 * 0.32 position variance 0.06 0.05 0.04 0.04 Sum of mean 63.45 57.96 52.76 67.37 4.29 * 0.17 st-wh angle variance 286.33 384.79 117.79 536.85 Mean st- mean 19.93 18.02 17.81 19.52 21.35 0.56 wh angle variance 1.60 1.71 1.27 3.35 Variance mean 40.28 29.21 27.35 36.31 9.35 0.34 st-wh angle variance 162.54 97.70 29.77 270.49 St-wh mean 6.50 5.48 5.41 5.78 1.18 0.01 angular vel variance 3.39 6.42 3.94 4.72

(44)

Appendix

Page 7 (19)

A E F F-value Omega2 Speed mean 92.11 71.90 71.60 75.05 0.89 variance 93.48 67.08 35.33 Lateral mean -0.60 -1.07 0.94 5.24 0.32 position variance 0.19 0.02 0.02 Sum of mean 72.29 61.59 60.35 2.53 0.15 st-wh angle variance 380.99 95.17 87.94 Mean st- mean 18.24 16.67 16.80 11.61 * 0.54 wh angle variance 6.29 3.59 2.39 Variance mean 51.61 38.37 44.23 7.13 * 0.41 st-wh angle variance 122.56 171.67 313.75 St-wh mean 10.43 6.68 6.39 12.31 0.56 angular vel variance 14.38 4.23 2.53

Sum of mean 6.19 5.14 5.31 3.06 0.19 yaw angle variance 1.17 0.81 0.58

Mean mean 0.86 0.54 0.94 16.92 0.64 yaw angle variance 0.03 0.02 0.04

Variance mean 0.62 0.18 0.33 3.74 0.23 of yaw ang variance 0.23 0.01 0.03

Lateral mean 1.79 1.12 1.09 51.63 0.85 accel variance 0.16 0.07 0.03

A E F F-Value Omega2 Speed mean 86.90 74.44 71.64 17.64 0.65 variance 124.51 73.06 25.93 Lateral mean 1.04 -1.09 1.04 0.25 -0.09 position variance 0.08 0.06 0.02 Sum of mean 76.59 69.23 79.93 2.55 0.15 st-wh angle variance 427.77 446.51 673.33 Mean st- mean 17.50 17.22 16.81 0.45 0.07 wh angle variance 6.40 3.87 2.19 Variance mean 56.51 41.00 55.39 6.61 * 0.38 st-wh angle variance 220.83 136.26 289.83 St-wh mean 9.99 7.18 9.12 3.08 0.19 angular vel variance 6.28 9.73 17.44

Sum of mean 7.14 5.19 7.50 9.17 * 0.48 yaw angle variance 7.02 1.92 2.17

Mean mean 0.70 0.57 0.93 27.85 0.75 aw angle variance 0.02 0.02 0.02

Variance mean 0.38 0.17 0.38 4.32 * 0.27 of yaw ang variance 0.06 0.01 0.04

Lateral mean 1.62 1.20 1.09 1 1.70 0.54 accel variance 0.21 0.07 0.02

(45)

Appendix

Page 8 (19)

A E F F-value Omega2 Speed mean 96.62 77.41 79.15 34.75 0.79 variance 108.06 46.83 67.36 Lateral mean 1.06 -1.11 -1.15 0.58 0.05 position variance 0.06 0.02 0.05 Sum of mean 67.23 67.42 70.95 0.03 -0.12 st-wh angle variance 147.94 2118.80 663.84 Mean st- mean 16.53 14.81 16.05 3.12 0.19 wh angle variance 6.23 1.66 4.57 Variance mean 81.45 64.35 84.84 2.48 0.14 st-wh angle variance 316.41 523.09 670.02 St-wh mean 10.22 9.12 8.71 0.09 -0. 11 angular vel variance 8.79 77.60 41.80

Sum of mean 6.71 5.26 6.72 0.94 -0.01 yaw angle variance 3.97 7.63 3.57

Variance mean 0.52 0.26 0.53 2.36 0.13 of yaw ang variance 0.09 0.04 0.12

A E F F-value Omega2 Speed mean 91.16 75.55 75.03 24.95 0.73 variance 111.05 56.02 39.92 Lateral mean -1.06 1.14 1.17 1.02 0.00 position variance 0.07 0.01 0.03 Sum of mean 62.26 54.05 78.78 3.41 0.21 st-wh angle variance 287.92 353.18 1706.36 Mean st- mean 15.12 14.50 14.67 0.45 0.07 wh angle variance 3.56 2.33 2.87 Variance mean 74.24 59.28 82.73 6.39 * 0.37 st-wh angle variance 179.04 309.74 1043.83 St-wh mean 8.92 5.84 9.05 2.62 0.15 angular vel variance 17.63 4.20 25.06

Sum of mean 5.79 4.69 6.51 2.88 0.17 yaw angle variance 4.00 3.51 6.62

Variance mean 0.38 0.24 0.44 5.91 * 0.35 of yaw ang variance 0.05 0.02 0.04

Driver behaviour on winter roads : A driving simulator study

VTI rapport 419A - 1997

Driver Behaviour on

Winter Roads

A Driving Simulator Study

Carl-Gustaf Wallman

Swedish National Roadand

'Transport Research Institute

Swedish National Road and

Project

Preface

Contents

Summary

1 Background

A

2 Problem

3 Hypothesis

4 Definitions and Measures

5 Method

880

V

440

560

\_/

\

/

_C_________________-37. _______.

To

6 Results

If:

7000

7100 "'50 7200

_

90

A

E

B

100

-A

90

80

-E

70

F

.51

H

A

M

45 1

H

10800

10400 10450 10500

f\

-45J

I

W 10850' 70900

7 Other Studies

8 Conclusions and Discussion

9 Future Research

10 References

Statistical Analysis

Tables over F-values, level of significance 0.05

_C___________-37. _.