Working Paper R 3.2.1
Evaluation of in-car speed limiters:
Simulator study
Public
MASTER
Contract No RO-96-SC.202
Project
Coordinator:
Partners:
Associate
partners:
VTT Communities & Infrastructure (VTT, Finland)
FACTUM Chaloupka, Praschl & Risser OHG (FACTUM, Austria)
University of Leeds - Institute for Transport Studies (ITS, U.K.)
KTI Institute for Transport Sciences Ltd (KTI, Hungary)
University of Lund - Department of Planning and Engineering (LU, Sweden)
TNO Human Factors Research Institute (TNO, the Netherlands)
Transport Research Laboratory (TRL, U.K.)
University College London - Centre for Transport Studies (UCL, U.K.)
INTRA S.L. (Spain)
TRANS-POR (Portugal)
SWOV Institute for Road Safety Research (the Netherlands)
Swedish Road & Transport Research Institute (VTI, Sweden)
Author:
Samantha Comte,
University of Leeds, U.K.
Date:
May 1998
PROJECT FUNDED BY THE EUROPEAN
COMMISSION UNDER THE TRANSPORT
RTD PROGRAMME OF THE
4th FRAMEWORK PROGRAMME
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
Summary
The European MASTER project (MAnaging Speeds of Traffic on European Roads) aims to
provide recommendations for speed management strategies and policies and develop guidelines
for the development of innovative speed management tools. This document describes a driving
simulator experiment carried out within Workpackage 3.2 (Evaluation of in-car speed limiters).
The University of Leeds Advanced Driving Simulator was used to test two speed control
systems against an advisory system and a baseline control (no system). The speed control
systems were both designed to prohibit exceeding the external speed limit; one of the systems
additionally applied further speed reduction in hazardous situations, such as sharp horizontal
curvature and poor weather conditions. The advisory system provided information to the
driver regarding appropriate speeds via an in-car display. A road environment incorporating
urban, rural and motorway scenarios allowed the comparison of the systems across road types.
Driver behaviour under the two control systems was compared to that in the advisory and
baseline conditions. Behavioural parameters measured include speed and its derivatives, time
headway, overtaking manoeuvres, traffic light violations and collision measures. Subjective
measures of workload were taken to monitor any possible underload or overload effects, and
an acceptability questionnaire was administered to ascertain driver opinion about the systems.
Results indicate that there are safety benefits of control systems including a reduction of
maximum speed, speed variance and inappropriate speed at hazardous locations. In addition it
was found that the advisory system performed well, especially where the driver could perceive
the relevance of that information. However, there were observed secondary effects of the
speed control system which may compromise any safety benefits. Such effects included a
higher incidence of short time headways, delayed braking and a higher incidence of collisions.
Subjective mental workload scores did not differ between the conditions, but it was shown that
drivers found the advisory system more acceptable than the control systems.
2
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
CONTENTS
1. INTRODUCTION............................................................................................................. 4
1.1 Previous research ........................................................................................................ 4
1.2 System selection.......................................................................................................... 8
2. METHOD ......................................................................................................................... 9
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
Experimental design .................................................................................................... 9
Participants ................................................................................................................. 9
Procedure.................................................................................................................... 9
Leeds driving simulator ............................................................................................... 9
Simulated road network ............................................................................................ 10
System operation....................................................................................................... 10
Dependent measures.................................................................................................. 11
Data analysis ............................................................................................................. 11
3. RESULTS ....................................................................................................................... 12
3.1
3.2
3.3
3.4
3.5
3.6
3.7
Driver speed profiles ................................................................................................. 12
Inappropriate speed................................................................................................... 18
Traffic light violation................................................................................................. 19
Following behaviour.................................................................................................. 20
Collision measures..................................................................................................... 21
Travel time................................................................................................................ 22
Subjective measures .................................................................................................. 22
3.7.1 Mental Workload ............................................................................................ 22
3.7.2 Acceptability ................................................................................................... 22
4. DISCUSSION AND CONCLUSIONS............................................................................ 25
5. REFERENCES................................................................................................................ 27
3
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
1. INTRODUCTION
The aim of MASTER (MAnaging Speeds of Traffic on European Roads) is to provide
recommendations for speed management strategies and policies and develop guidelines for the
development of innovative speed management tools. The project consists of three research
areas:
• Acceptable ranges of speed
• Influences on speed behaviour
• Speed management tools and strategies
Workpackage 3.2 forms part of the third research area and aims to identify which Advanced
Transport Telematics (ATT) systems for speed management have been developed or are being
contemplated, examine their market potential and test the most promising systems in terms of
likely safety benefits. This report details a simulator experiment carried out to evaluate the
effects of in-car speed limiters on driver behaviour. A field trial with an instrumented car also
took place in parallel and is reported in Deliverable 11.
1.1 PREVIOUS RESEARCH
The first field trial with a speed limiter in passenger cars was carried out in France (Saad and
Malaterre, 1982). The driver could set a speed limit, which could not be exceeded unless the
driver disengaged the equipment by a separate operation. It was found that the test drivers
adapted their speed in relation to the traffic situation around them, giving rise to small changes
in speed around the speed limit. In order to “float with” the other cars, drivers had to
disengage the speed-limiter quite frequently which was reported to be physically tiring. They
also found keeping to the speed limit on 60 km/h and 80 km/h roads difficult. On 90 km/h
roads the use of limiter depended on the traffic volume, such that a higher traffic volume
resulted in frequent changes in speed and thus less use of the limiter. On roads with a speed
limit of 110 km/h and 130 km/h where monotonous driving was more common, the limiter was
used more frequently. Most drivers set the top speed on the limiter significantly above the
speed limit and it was found that as the speed limit on a road decreased, the difference between
the speed set on the limiter and the road speed limit increased. On shorter stretches of road
with a lower speed limit, e.g. through villages or curves, the drivers did not adjust the speed
set on the limiter. In addition, the knowledge that the speed limited vehicle had less available
acceleration, meant that drivers in some circumstances chose not to perform an overtaking
manoeuvre.
A field study in Sweden (Persson, Towliat, Almqvist, Risser and Magdeburg, 1993) showed
that mean speed decreased on links by between 2% and 8% with the speed limiter but there
was a slight tendency to compensate for the low speeds on links by driving faster (by 2-3 km/h)
through the junctions. Data from behavioural observations showed a clear increase of the
proportion of correctly kept distance to the car ahead (even on sites with speeds lower than the
actual speed limit). On the other hand, there was a slight increase of incorrect behaviour
4
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
towards other road-users at junctions. Most of the drivers generally displayed positive
attitudes towards the speed limiter but did sometimes experience feelings of impatience.
Eighty-five % of all drivers reported positive attitudes with regards to safety. The most often
mentioned advantage for drivers was smoother rhythm in traffic and better preparedness for
unprotected road-users. The most commonly mentioned disadvantage for drivers was that it
would be impossible to accelerate and sometimes exceed speed limits where necessary.
Behavioural adaptation with speed limiters at junctions was also demonstrated in a simulator
study (Comte, 1996). A previous experiment carried out on the University of Leeds Advanced
Driving Simulator evaluated the effects of speed limiters on driver behaviour. Safety was
measured in terms of following behaviour, gap acceptance, traffic violations, and subjective
mental workload was recorded using the NASA-RTLX. It was found that although safety
benefits were observed in terms of lower speeds, longer headways and fewer traffic light
violations, drivers compensated for loss of time by exhibiting riskier gap acceptance behaviour
and delayed braking behaviour. When speed was limited, drivers’ self-reports indicated that
their driving performance improved and less physical effort was required, but that they also
experienced increases in feelings of frustration and time pressure.
A Finnish traffic simulation study on the effects of compulsory speed limiters on heavy vehicles
showed that traffic safety improved due to a decrease in accidents (Kulmala and Beilinson,
1993). The decreases in travel speeds were larger for heavy vehicles (at the highest between
2.2 and 2.9 km/h at free flow) than for cars. The effects on the standard deviation of speeds
were quite small.
Davidsson (1995) simulated a fully implemented dynamic speed adaptation system in a central
built-up area in Sweden. The maximum allowed speed through junctions was limited to 25
km/h. The speed variation decreased from 8.4 km/h to 4.2 km/h, while the mean speeds, in
spite of lower maximal speeds, increased slightly. This is due to the fact that the vehicles
drove in platoons in a larger extent and passed the junctions more smoothly. It was concluded
that the system would give travel time savings up to 25%.
In an ongoing project in the Swedish town of Eslöv a number of company cars are being
equipped with speed limiters. On all entry roads transponders (radio transmitters) are mounted
on the 50 km/h speed signs. When the equipped cars pass the transponders the speed limiter is
automatically set on the 50 km/h speed limit and when the cars leave the town the speed limiter
is deactivated. Results showed there to be speed reductions and decreased speed distributions,
and driver acceptability increased after the trials (Almqvist and Nygård, 1997).
In July 1996, the Ministry of Transport, Public Works and Water Management in The
Netherlands, presented a new multilayer programme for road safety including a pilot study to
assess the feasibility of using intelligent speed adaptors (ISA). The project will be carried out
on a new housing estate in Tilburg to be built in 1998, the main aim of which will be to
increase public acceptance and support for such a device. The study hopes to deal with
aspects such as technical feasibility, integration, legal framework and organisational
consequences. Three technological variants have been chosen for detailed examination :
5
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
• direct adjustments to the engine to make it impossible to drive faster than the local speed
limit.
• increased counterpressure on the accelerator pedal when the speed limit is exceeded
• an auditory signal when the speed limit is exceeded.
Researchers have attempted to calculate the likely savings in terms of accident reduction that
speed limiting systems might have. For example Várhelyi (1996) analysed the occurrence of
accidents under differing road and lighting conditions and concluded that a speed limiting
system would lower the number of police reported injury-accidents by between 19% and 34%.
The author also states that this is a conservative estimation, because it does not include
accidents that occur under impaired visibility such as those in fog or on sharp curves. It is
suggested that a more optimistic estimate is a reduction in accidents between 24% and 42%.
U.K. researchers, Perrett and Stevens (1996) suggest the key costs and benefits from
automatic speed control in the U.K. would be as described in Table 1. They go on to suggest
likely accident savings if such an implementation were to be introduced by using a test scenario
based on mandatory automatic speed control implemented on all roads in Great Britain. The
envisaged system would consist of a transceiver linked to an in-vehicle display and speed
governor, which is operated by the engine management system. They postulate that speed
related accidents would have to be reduced by 50% in order for the benefits to match the costs.
However, some of their assumptions, e.g. on the number of relevant accidents and on
infrastructure costs, are questionable. In addition, they did not conduct sensitivity analyses.
Table 1. Key costs and benefits from automatic speed control (Perrett and Stevens 1996)
Key costs and benefits
Justification
Public investment costs for
transponders: £145.8m
Hypothesis that speed limits change every 5 km
Public investment cost: £250/new
car
Operating cost: 1% of investment
cost/year
Hypothesis based on cost of existing systems
Accidents reduced by 16500/year
Hypothesis based on PROMETHEUS research which
suggested that 16% of accidents would be avoided, and that
half of the potential would be realised by this application, due
to the lack of facility to vary speed limits dynamically
Estimate based on data on fuel consumption at different
speeds, vehicle speed survey, and distance travelled on
motorways
Fuel use on motorway journeys
reduced by 89.5m litres/year
Enforcement costs reduced by
£50m/year
Infrastructure investment cost
reduced by £25m/year
Hypothesis that a nominal saving is made when
implementation reaches 90% of vehicles
Hypothesis that a nominal saving is made when
implementation reaches 90% of vehicles
6
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
When introducing any new system to the market it is of importance to understand current
public acceptability to the system, as this could have a large impact on its effectiveness. Several
studies with different systems for speed adaptation, based on new technology, show that the
acceptance of such systems is fairly good. The SARTRE study (Dahlstedt, 1994) showed that
devices to control the speed of cars “when the driver is free to turn it on or off” were favoured
by about 46% of the respondents, while about 42% were against. According to Nilsson (1996)
it is reasonable to assume that systems which increase safety and efficiency, which are reliable
and user-friendly and which are not used for control of the driver will be gradually accepted by
the drivers.
Comte, Wardman and Whelan (1997) conducted acceptability exercises regarding vehicle
speed limiters. A Stated Preference exercise compared traditional methods of enforcement and
traffic calming to vehicle speed control. In terms of the acceptability of speed reduction
methods, measures relating to stricter enforcement were regarded to be more acceptable than
methods which might reduce speeds below current legal levels. The ordering of acceptability
was speed cameras, speed limiters and traffic calming. With regard to the effectiveness of
different measures in reducing speeds, speed limiters were, as might be expected, regarded to
be the most effective followed by speed cameras and traffic calming. Focus groups were also
carried out with road safety experts and members of the general public. Speed control of some
sort was welcomed, but the idea of control being external to the driver was not well received,
and it was commented that public acceptability was likely to be low at first. However, it was
thought that if speed control was implemented it should be a mandatory system, whose launch
should be combined with a positive marketing and lowered costs in terms of, for example,
insurance premiums. It was also suggested that if the technology was such that speed control
could be introduced then drivers should also receive for example route guidance information in
order to increase acceptability of speed control.
Although systems such a speed limiters have the potential to improve driver safety and
comfort, they have the disadvantage that they change the nature of the driver task. Ultimately,
speed limiters automate some components of the driving task, i.e they decrease the need for
the driver to monitor their speedometer in order to keep to the speed limit and/or to avoid
citations. So although speed limiters offer the benefits of reduced maximum speeds and
possibly decreased speed variance, any evaluation of such a system should include the
measurement of any secondary effects which may degrade the value of the safety benefits. For
example automating parts of the driving task could affect driver workload. If additional
information is provided via an in-car display, required cognitive functioning may increase. Such
increases may lead to driver overload which could be potentially hazardous in terms of the
ability to optimally carry out the driving task. Alternatively, automation of part of the driving
task may remove drivers out-of-the-loop and lead to driver underload and hence loss of
situation awareness (SA). Situational awareness refers to an awareness of environmental
information relevant to successful task performance, the meaning and context of that
information and the predictive future state of the environmental conditions. A loss of SA may
mean that drivers are less responsive to critical incidents. An example of such loss of SA was
demonstrated in a field trial evaluating an Adaptive Cruise Control (ACC) system in terms of
performance measures (Ward, Fairclough and Humphreys, 1995). Adaptive Cruise Control
7
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
(ACC) automates the task of maintaining vehicle speed conditional to a minimum following
distance. The system instigates braking or acceleration dependent on whether the detected
headway is recognised to be smaller than a set criterion. They found that although ACC
resulted in the lower incidence of short following distances, there were more errors observed in
terms of proper lane keeping and yielding to other traffic. A similar result was found in a
simulator based study examining the effect of automation on the driving task in terms of
reduced arousal (Richardson, Ward, Fairclough and Graham, 1996). Drivers encountered a
queue of stationary traffic at the end of a one hour driving session. It was reported that the
minimum time-to-collision in responding to this event was significantly shorter with ACC than
for unassisted driving. The authors of these two studies suggest that drivers either may have
become complacent and reliant on the system, or that the additional demands of the ACC
interface distracted drivers from the primary driving task.
1.2 SYSTEM SELECTION
The study reported here evaluated three systems designed to reduce driver speed. The first of
these systems, is termed the Advisory system. It has been suggested (e.g. De Waard, Jessurun,
Steyvers, Raggatt and Brookhuis, 1995) that speeding may be due to general unawareness of
the speed limit. The advisory system provided drivers with a continuous reminder of the
external speed limit, via a visual display, without exerting any control over the car. Thus the
driver remained in-the-loop, but could be considered as more informed of the external
conditions. The second system, the Fixed system, limited the maximum speed of the car to the
speed limit of the road along which the car was travelling. Thus the driver was not able to
exceed the speed limit. The final system, the Dynamic system, operated in the same way as the
fixed system and additionally lowered speed further in poor road and weather conditions.
These three systems represent a hierarchy of increasing costs and control over the driver.
Driver behaviour under these three systems was compared to that in a baseline condition where
drivers were unassisted. Behavioural parameters measured include speed and its derivatives,
time headway, overtaking manoeuvres, traffic light violations and collision occurrence.
Subjective measures of workload were taken to monitor any possible underload or overload
effects, and an acceptability questionnaire was administered to ascertain driver opinion about
the systems.
It was hypothesised that safety benefits would occur using the systems including reductions in
excessive speeds and speed variability. It was additionally hypothesised that these safety
benefits may be counteracted by secondary effects of loss of awareness, complacency and
behavioural adaptation.
8
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
2. METHOD
2.1 EXPERIMENTAL DESIGN
Three speed management systems were evaluated against a baseline condition. A between
subjects design was employed to provide the opportunity of presenting drivers with a critical
incident at the end of the road network. A within subjects design would have meant the drivers
encountering critical incidents on more than one occasion and thus learning may have
occurred. Participants were matched for average annual mileage between the four conditions.
Participants were allocated to one of four groups corresponding to the three systems and one
control condition and each participant drove the road network of 25km once.
2.2 PARTICIPANTS
A total of 60 participants took part in the experiment. All participants possessed a full, clean
driver’s licence. Participants who suffered simulator sickness during the trial were excluded
from the analyses.
2.3 PROCEDURE
Participants were requested to complete a practice drive on the simulator in order that they
familiarise themselves with the controls of the car. Following this, they completed the
experimental route and were then asked to fill out the NASA RTLX, a standard measure of
mental workload. This requires subjects to rate the task they have just completed in terms of
mental demand, physical demand, time pressure, performance, effort and frustration level.
Each of these items were represented by a bipolar scale; subjects placed a line on the scale
between the two extremes of the item to indicate the strength of the attribute. The individual
scales are then averaged to give a total workload score (see Appendix A). In addition, where
appropriate, participants’ opinions towards the systems were obtained. An acceptability scale
and a questionnaire were administered for this purpose (see Appendix A). The total time to
complete the experiment was approximately one hour.
2.4 LEEDS DRIVING SIMULATOR
The experiments were carried out on the University of Leeds Advanced Driving Simulator.
This is a fixed base simulator presenting a 120º forward view and 50º rear view. The system
features a fully interactive Silicon Graphics (Onyx RE²) driving simulator with a six degree of
freedom vehicle model. A servo motor linked to the steering mechanism provides control over
handling torque and speed and digitised samples of engine, wind, road and other vehicles are
provided. Photo-realistic scene texturing allows presentation of various road types and
features.
9
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
A recent study (Carsten, Groeger, Blana and Jamson (1997) evaluated the behavioural validity
of the Leeds Driving Simulator. The results show that overall there was a broad
correspondence between driving in the simulator and the behaviour of real-world traffic. With
regard to speed, the effects of road width, curvature, direction of curve and sequence between
road sections were reproduced on the simulator, and there were very high correlations between
speed along the real road and speeds in the simulator. With regard to lateral position, the
correlations were less satisfactory, but this is hardly surprising in view of the absence of a
motion base in the simulator. It should therefore be possible to draw conclusions based on
speed behaviour in the simulator.
2.5 SIMULATED ROAD NETWORK
The simulated road network incorporated urban, rural and motorway environments to allow
testing of the systems in a variety of external speed limits and scenarios. Urban environments
(48km/h speed limit) were created primarily to allow participants to interact with other traffic;
rural environments (96km/h speed limit) allowed the investigation of sub-standard curve
negotiation; the motorway scenario (112 km/h speed limit) presented the opportunity to study
possible driver underload effects in terms of reduced vigilance. Traffic violations and unsafe
driver behaviour were recorded. This was achieved by providing car following scenarios,
critical decisions at traffic lights and overtaking opportunities. There was other traffic present
in the database, however, apart from the following tasks, participants encountered free-flowing
conditions. In addition a critical incident was placed at the end of the network, whereby
participants encountered a stationary row of cars on the motorway. The stationary cars were
partially obscured by dense fog, requiring rapid and harsh braking to avoid colliding with
them.
2.6 SYSTEM OPERATION
Three systems were tested under identical traffic and road environment conditions.
• Advisory system This system provided drivers with a continual reminder of the external
speed limit. In addition advisory speeds for any hazardous conditions ahead were also
displayed. In this experiment drivers were warned they were approaching a sub-standard
horizontal curve and the appropriate advisory speed was displayed by the system. In
addition, a message warning of fog was displayed and again the appropriate advisory speed
was displayed by the system.
• Fixed system This system automatically limited the car to the external speed limit. If the
driver travels from a higher speed limit to a lower one, the system automatically reduces the
speed of the simulator in readiness for the lower speed limit. Thus the driver will be
travelling at the speed limit as they pass the speed limit sign (as is required under traffic
law). When the driver is travelling from a low speed limit to a higher one, then the system
10
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
allows acceleration only after the new speed limit is in operation. See Appendix B for more
details of the system operation.
• Dynamic system In addition to automatically limiting the speed of the car to the external
speed limit, this system also further reduces speed in hazardous situations. In this
experiment drivers were warned they were approaching a sub-standard horizontal curve and
the appropriate advisory speed was displayed by the system. The system was activated and
drivers could be decelerated if necessary to this advisory speed. In addition, a message
warning of fog was displayed and subsequent deceleration to the advisory speed then
occurred.
The advice was provided via an in-car display situated on the dash board (See Appendix B).
2.7 DEPENDENT MEASURES
Speed measurements were taken every 10 metres throughout the whole journey. In addition,
indices of safety critical behaviour such as minimum time to collision in following tasks and the
incidence of overtaking manoeuvres were recorded. Traffic light violations, speed violations
and curve negotiation behaviour were also noted.
Data were also collected relating to driver reaction to the critical event situated at the end of
the road network. Firstly, the occurrence of collisions with the lead vehicle were recorded. A
collision was defined as the simulator car coming into the contact with the rear of the lead
vehicle. If drivers did collide with the lead vehicle, the speed on impact was also recorded as an
indication of collision severity. Headway to the lead vehicle was measured both at the time of
initial braking and at the instant the brake pedal was fully depressed.
2.8 DATA ANALYSIS
One-way ANOVAs were used where assumptions of analyses of variance were not violated
and post-hoc multiple comparison tests (Tukey’s HSD) were performed in order to determine
whether the speed management had differential effects on driver behaviour. Where the data
violated the assumptions of normal distribution and equal sample variances, nonparametric
Kruskal-Wallis H tests were used. This tests the hypothesis that the treatment medians do not
differ significantly. Details of the statistical analysis can be found in Appendix C.
11
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
3. RESULTS
3.1 DRIVER SPEED PROFILES
Six sections of the road network were extracted for the analysis of driver speed profiles. The
sections covered all road types and speed was recorded every 10m where the driver was under
free flowing traffic conditions.
Mean speed for these sections under each of the systems were calculated as shown in Table 2.
Table 2: Mean speed through road sections
Mean speed [SD]
Section
(speed limit, km/h)
Baseline
Advisory
Fixed
Dynamic
Village 1 (48)
Village 2 (48)
Village 3 (48)
Single carriageway (96)
55.86 [7.62]
46.66 [5.37]
56.66 [5.34]
72.60 [10.63]
52.32 [5.32]
44.15 [3.52]
53.61 [4.67]
71.46 [7.67]
46.75 [1.69]
43.50 [2.69]
47.60 [0.68]
77.25 [10.93]
45.72 [2.84]
42.63 [3.48]
46.71 [1.36]
72.45 [10.45]
Dual carriageway(112)
Motorway (112)
104.31 [9.08]
88.50 [13.17]
107.64 [9.92]
79.66 [10.07]
107.18 [5.82]
96.03 [8.57]
106.57 [8.02]
76.78 [4.54]
A comparison of sample variances showed that some sections demonstrated unequal variances.
Thus one-way analysis of variances or Kruskal-Wallis tests were carried out appropriately on
the separate sections. They revealed significant differences between the conditions for the
village sections (F[3,56]; p < 0.05) and the motorway section (F[3,56]; p < 0.01). Post-hoc
analyses showed that mean speeds in the villages were higher in both the baseline and advisory
conditions compared to the fixed and dynamic conditions. However in the motorway section,
mean speed was higher in the fixed condition than in either the advisory and dynamic
conditions (F[3,56]; p< 0.01).
Driver speed profiles in urban areas are shown in Figures 1-3. In Figure 1, the deceleration
profiles on approach to the village are shown for each of the conditions. At the change in speed
limit (denoted by the 48 km/h speed limit sign), drivers who were not speed-controlled had
only reduced their speed by approximately 15 km/h (as opposed to the appropriate reduction
of approximately 35 km/h). Those in the advisory condition appear to have reduced their speed
only slightly more than those in the baseline condition, but this was probably due to the fact
that changes in the advisory speed were displayed at the point where drivers passed the speed
limit sign on the road.
12
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
100
Speed (km/h)
Baseline
90
Advisory
Fixed
80
Dynamic
Speed Limit
Speed limiter
activated
70
60
50
40
48
600
550
500
450
400
350
300
250
200
150
100
50
0
30
Distance along section (m)
Figure 1: Speed profile through Village 1
Figure 1 also shows there is a difference in speed passing the traffic lights. These traffic lights
were included to evaluate driver behaviour on their approach to them. It was hypothesised that
drivers using a speed control system would maintain their maximum speed at locations where
drivers not under control would naturally decelerate. Traffic lights are an example of a
situation where some deceleration might occur in anticipation of them turning to red. Indeed it
can be seen from Figure 1 that some deceleration does occur in the advisory and baseline
conditions before the traffic lights. However, it also appears that drivers in the speed controlled
conditions also engage in a slight amount of deceleration. A spot speed was taken at the traffic
lights to discover any differences in speeds between the conditions. The data revealed to have
unequal sample variances (Levene p<0.001), and were thus analysed with a nonparametric
Kruskal-Wallis test. There was found to be a difference between the conditions (χ²= 11.16, p<
0.01). Multiple comparison post-hoc LSD (least significant difference) tests at 0.05
significance level revealed the speed passing traffic lights was higher in the baseline condition
than in any of the other conditions.
Figure 2 shows the driver speed profiles through the second village in the road network. This
was a relatively long village and only two sections of interest are shown. The first is a straight
section of urban road which was located directly after subjects had been forced to follow a
slow moving vehicle. The profile shows that drivers in the baseline condition attained speeds of
approximately 55km/h, with those in the advisory condition achieving slightly less.
13
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
65
60
Straight section
Area with
pedestrians
Speed (km/h)
55
50
45
Baseline
40
Advisory
35
Fixed
30
Dynamic
Speed Limit
900
800
700
600
500
400
300
200
100
0
25
Distance along section (m)
Figure 2 : Speed profile through Village 2
The second section of the profile in Figure 2 depicts a location where pedestrians were situated
and parked cars effectively narrowed the available road width. It can be seen that drivers in all
the experimental conditions travelled at or slightly under the speed limit. Those in the baseline
condition travelled above the speed limit. Thus it seems that the systems were effective in that
they reduce ‘normal’ driving speeds through built-up areas, and it appears the advisory system
is sometimes effective in its role of reminding the driver of the appropriate speed at which to
travel.
14
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
100
Baseline
Speed limiter
activated
90
Advisory
Fixed
Speed (km/h)
80
Dynamic
Speed Limit
70
Derestriction
sign in sight
60
50
40
48
1300
1200
1100
1000
Distance along section (m)
900
800
700
600
500
400
300
200
100
0
30
Figure 3: Speed profile through Village 3
The speed profile in Figure 3 is for a village situated in the road network after a rural section
(96 km/h). It demonstrates the real-life speeding phenomena of ‘speed adaptation’ which can
occur in rural villages. Drivers who have been driving at a high speed for an extended period
may become habituated to the speed and overestimate the degree to which they are lowering
their speed. At the speed limit sign, drivers in the baseline condition are still travelling
approximately 15 km/h above the speed limit. Interestingly again, the advisory system seems to
having an effect on speed even though the change in speed limit is not displayed until they pass
the speed limit sign. Driving speeds through the village in the baseline condition never descend
to the same level as in the other three conditions. Drivers in the baseline and advisory
conditions also begin anticipatory acceleration before the speed limit increases.
Figure 4 shows the profile of speed on the motorway section in foggy conditions. This
motorway section was placed at the end of the road network and the speed profiles show quite
clearly that drivers with the fixed speed limiter exceed speeds of even those with no speed
control at all.
15
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
120
110
Speed (km/h)
100
90
80
Baseline
70
Advisory
Fixed
Fog warning
60
Dynamic
Speed Limit
1100
1000
900
800
700
600
500
400
300
200
100
0
50
Distance along section (m)
Figure 4: Speed profile on motorway section
Only drivers in the advisory condition appear to decrease their speeds after receiving the fog
warningbe on the screen, soon after entering the section. There are two possible explanations
for such behaviour. Firstly drivers with the fixed speed limiter may have become reliant on the
system to maintain them at a safe speed. Drivers appear to be increasing speed towards the
speed limit because the system allows them to. Secondly, the difference in speed may be due to
behavioural adaptational effects: drivers have been travelling at the speed limit for the last 25
minutes and see this location as an opportunity to make up for lost time or increase their sense
of thrill.
As well as recording mean speed through the road sections, standard deviation of speed was
calculated as a measure of speed stability. From Table 3 it can be seen that there are
differences in speed variation between conditions where drivers are speed controlled and those
where they are not controlled.
16
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
Table 3: Mean standard deviation of speed through road sections
Standard deviation speed [SD]
Section
(speed limit, km/h)
Village 1 (48)
Baseline
Advisory
Fixed
Dynamic
7.50 [3.12]
6.43 [2.94]
1.10 [1.10]
1.57 [1.23]
Village 2 (48)
8.20 [2.83]
7.74 [2.39]
5.26 [1.94]
5.53 [2.23]
Village 3 (48)
Single carriageway (96)
6.62 [2.52]
10.68 [3.15]
6.95 [2.66]
13.12 [2.64]
0.89 [1.03]
11.40 [3.29]
2.32 [2.33]
11.50 [3.8]
Dual carriageway(112)
Motorway (112)
7.46 [3.14]
9.69 [4.49]
7.34 [2.00]
5.85 [2.66]
3.97 [4.09]
9.29 [3.25]
3.82 [3.90]
3.13 [1.34]
As in the case for mean speeds, a comparison of sample variances showed that some sections
demonstrated unequal variances. Thus one-way analysis of variances or Kruskal-Wallis tests
were carried out appropriately on the separate sections. There were found to be significant
difference for all road types, (F[3,56]; p<0.01), apart from the rural single carriageway, where
speed was expected to be variable in all conditions due to the curvature of the road. Speed
variation is reduced in villages under speed control. In the motorway section however,
although speed variance is low in the Dynamic condition (due to the restriction to 80 km/h), it
can be seen that speed variance is virtually identical in the fixed and baseline condition. The
fixed speed limiter performs significantly even more poorly in terms of speed variation than the
advisory system (F[3,56]; p<0.01).
Table 4 : Maximum speed through road sections
Baseline
Maximum speed [SD]
Advisory
Fixed
Section
(speed limit, km/h)
Village 1 (48)
Dynamic
72.91 [11.70]
66.89 [8.59]
48.14 [0.11]
47.52 [1.84]
Village 2 (48)
60.55 [7.68]
58.78 [8.21]
48.23 [0.14]
48.26 [0.11]
Village 3 (48)
Single carriageway (96)
Dual carriageway(112)
71.86 [9.87]
90.08 [12.76]
118.21 [8.47]
71.76 [5.95]
92.94 [9.36]
122.66 [8.42]
50.53 [3.71]
92.31 [7.57]
111.59 [1.76]
51.33 [5.73]
88.94 [9.01]
111.53 [2.13]
Motorway (112)
105.81 [16.9]
90.87 [11.31]
107.31 [7.76]
85.18 [96.95]
Driver maximum speed was also recorded through each of the sections as shown in Table 4.
Analyses of variance and Kruskal-Wallis tests revealed significant difference between the
conditions on all of the sections apart from the rural single carriageway (p<0.001). Post-hoc
analysis indicate drivers without speed control reach significantly higher maximum speeds than
if they are speed controlled, in urban areas.
17
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
3.2 INAPPROPRIATE SPEED
Spot speeds were taken at potentially hazardous locations on the road network. These
locations were those where the dynamic system was imposing a speed that was below the
external speed limit, or advisory speed systems was displaying a reduced advisory speed. Of
major interest was whether the advisory system would be as effective as the dynamic system.
For example, entry speed to a sub-standard rural curve was recorded; here the dynamic speed
control automatically reduced driver speed to the recommended curve negotiation speed of 40
km/h. For drivers with the advisory system, a message saying “Sharp bends, slow down to 40
km/h” was shown on the in-car display. The mean curve entry speeds are shown in Table 5.
Table 5: Curve entry speeds (km/h)
Baseline
Advisory
Fixed
Dynamic
39.88
34.98
38.09
32.77
A significant difference was found between the conditions (F[3,56]; p<0.01). However this
difference existed only between the baseline and dynamic systems. Although the advisory
system appears to have reduced speeds to some extent, this difference did not reach statistical
significance.
Figure 5 shows the speed profiles of drivers as they approach this rural curve. Drivers appear
overall to be taking note of the advisory speed, although their deceleration profile is less steep
than in the dynamic condition. This seems to matter little however, as curve entry speed is
similar in these two conditions. Drivers in the fixed and baseline conditions begin their
deceleration towards the curve much later.
18
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
100
90
80
Speed (km/h)
70
60
Dynamic speed limiter
and advisory system
activated
50
Baseline
40
Advisory
30
Fixed
20
Dynamic
10
Speed Limit
Bend
entry
Advisory Speed
800
700
600
500
400
300
200
100
0
0
Distance along section (m)
Figure 5: Speed profile on approach to rural curve
Mean speed along the stretch of motorway where conditions were foggy were measured. The
values are shown in Table 6.
Table 6: Mean speeds in fog (km/h)
Baseline
Advisory
Fixed
Dynamic
91.65
79.74
97.74
76.34
It is expected that the dynamic condition produces lower speeds, as the system is activated. An
analysis of variance revealed there to be significant difference of mean driving speed between
the conditions (F[3,56]; p<0.01), and post-hoc tests showed the advisory system was as
effective as the dynamic condition in reducing speeds as compared to the fixed and baseline
conditions.
3.3 TRAFFIC LIGHT VIOLATION
A situation was created whereby drivers were forced to make a rapid stop/go decision at
oneset of traffic lights which turned from green to amber as they approached. The percentage
of violations in each condition is shown in Table 7.
19
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
Table 7 : Percentage of red light violations
Baseline
Advisory
Fixed
Dynamic
33 %
46 %
53 %
40 %
There was no statistically significant difference between the conditions.
3.4 FOLLOWING BEHAVIOUR
Percentage of trial duration
at <1 sec headway
The road network allowed the inclusion of four car following tasks. In two of these tasks (on
each of three road types) the driver was unable to overtake the car in front due to oncoming
traffic. This created a “boxed-in” situation which allowed the measurement of the time
headway distribution. The lead cars in these scenarios were travelling at a speed that was
constant and below the speed limit. Thus in the urban situation the lead car was travelling at 40
km/h, in the rural area at 65 km/h. Therefore, even if speed limited, it was possible for drivers
to adopt short headways if they wished to. The safety critical value of <1 second headway was
calculated for the car following task. There was found to be an increase in the amount of time
drivers adopted this headway in the speed limited conditions, particularly in the urban scenario,
see Figure 6.
12
10
urban
rural
8
6
4
2
0
Baseline
Advisory
Fixed
Dynamic
Figure 6: Percentage of trial duration spent at <1 second headway
The remaining two following events allowed drivers to overtake the lead car (which were
travelling at the same speed as in the “boxed-in” events. The number of attempted and
successful overtaking manoeuvres was recorded. No differences between the conditions were
found.
20
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
3.5 COLLISION MEASURES
In order to evaluate whether speed control might affect the occurrence of accidents, a potential
collision scenario was created at the end of the road network. A line of stationary traffic,
concealed in thick fog was placed across all three lanes of the motorway. This scenario was
placed at the end of the road network firstly for practical reasons but more importantly to
discover if driver vigilance, after approximately 25 minutes of driving under speed control was
reduced. A collision was defined as the subject’s car making contact with the rear of the
stationary cars. The distribution of collisions is shown in Table 8.
Table 8: Number of collisions
Baseline
Advisory
Fixed
Dynamic
1
0
3
0
It can be seen that the incidence of collisions was generally low. This was probably due to the
fact that, on detection of the thick fog, subjects naturally reduced their speed, thus permitting
them to brake timely and avoid the collision. However it can be seen from Table 8 that when
they did occur, it was generally in the Fixed condition. It could be suggested that because
driver speed impacts on the probability of a collision occurring, then the reason for the
increased incidence of collisions in the Fixed condition was due to the high mean speeds in this
condition on the motorway section (this can be seen quite clearly in Figure 4). Thus differences
in driver speed in each condition may have led to the different number of collisions between the
conditions. To examine this possibility, subject driver spot speeds at 150 m before the
stationary cars was compared between the conditions. Only a significant difference in driving
speed was found between the Baseline and Dynamic (where drivers were limited to the
advisory speed anyway). There were no differences between the Fixed, Advisory and baseline
conditions. Thus, even though velocity is an important factor in the likelihood of collisions, the
increase in number of crashes in the fixed condition was due to a confounding factor.
Collisions may have been more likely to arise in the fixed condition due to the nature of the
braking pattern drivers adopted. Time headway at maximum brake pressure was calculated and
are shown in Table 9.
Table 9: Time headway at maximum brake pressure (seconds)
Baseline
2.93
Advisory
2.82
Fixed
1.83
Dynamic
3.29
It appears that drivers in the fixed condition, applied their maximum braking later than in other
conditions. This might account for the increased number of collisions with this system, and
could be as a result of automation-induced complacency.
21
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
3.6 TRAVEL TIME
Travel time was recorded as the time taken to complete the simulator trial, shown in Table 10.
Table 10: Travel time in seconds
Baseline
Advisory
Fixed
Dynamic
1450.67
1466.39
1418.27
1500.32
There were found to be no significant impacts of any of the systems on travel time.
3.7 SUBJECTIVE MEASURES
3.7.1 Mental Workload
Driver mental workload was assessed subjectively using the NASA-RTLX. Subjective
workload scores were obtained for each condition in terms of mental demand, physical
demand, time pressure, performance, effort and frustration level. Each of these items were
represented by a bipolar scale; participants placed a line on the scale between the two extremes
of the item to indicate the strength of the attribute. The individual scales are then averaged to
give a total workload score. No significant differences were found between the conditions,
indicating no differences in workload between the four conditions.
3.7.2 Acceptability
Acceptability scores were taken both before and after participants encountered the system (the
system was described to them in written instructions beforehand). See Appendix A for a copy
of the acceptability questionnaire). This served to indicate any preconceptions drivers might
have about the systems under investigation and demonstrate whether use of the system
improved driver acceptability. An end-score for each subject on the two dimensions of
“usefulness” and “satisfying” was calculated for each system. An overall system score was
then obtained across participants. The results are shown in Figure 7 and Figure 8.
22
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
dynamic
pre
post
fixed
advisor
y
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
Acceptability rating
Figure 7 : Acceptability ratings on the dimension of “useful”
It appears that the experience of driving with the fixed and dynamic system has little effect on
whether respondents believed the system would be useful or not. However, acceptability
scores decreased after use of the advisory system.
post
dynamic
pre
fixed
advisory
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
Acceptability rating
0.4
0.6
0.8
1
Figure 8: Acceptability ratings on the dimension of “satisfying”
It seems that respondents had pre-expectations about the speed control systems in terms of
how satisfying they would be. They rated the advisory system as being much more acceptable
both before and after the drive.
As part of the acceptability exercise, several questions were also posed to respondents. The
results are shown in Tables 11-13.
23
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
Table 11: Response to question 1
Question 1: In your opinion, would this system make people drive more safely?
System
yes,
definitely
yes, probably
no, probably not
no, definitely
not
Total
Advisory
0
11
4
0
15
Fixed
Dynamic
3
2
6
8
6
3
0
2
15
15
Table 12: Response to question 2
Question 2: In your opinion, would this system make people commit less offences?
System
yes, probably
no, probably not
Advisory
yes,
definitely
0
Total
5
no, definitely
not
0
10
Fixed
Dynamic
4
6
8
6
3
1
0
2
15
15
15
Table 13: Response to question 3
Question 3: Would you have this system installed in your own car on a voluntary basis if it
cost in the region of £50?
System
yes, probably
no, probably not
Advisory
Fixed
yes,
definitely
2
0
Total
5
8
no, definitely
not
4
4
4
3
Dynamic
0
1
7
7
15
15
15
These questions reveal that drivers generally believe that speed control systems would
encourage drivers to drive more safely and commit less offences, however they are adverse to
actually owning one personally. Additional driver comments can be found in Appendix D.
24
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
4. DISCUSSION AND CONCLUSIONS
This simulator study evaluated automatic speed control in terms of driver behaviour
parameters, workload and acceptability. Two types of speed limiter were tested (fixed and
dynamic) against an advisory system and baseline condition. As hypothesised, speed limiter
systems were successful in reducing excessive speed, particularly in urban areas. The various
speed profiles showed that without speed control, drivers are susceptible to poor speed
adaptation. Drivers in both the baseline and advisory conditions do not decelerate to the speed
limit of the villages (48 km/h) as they pass the new speed limit, although the advisory system
does encourage drivers to reduce their speed further than in the baseline. This is known to be
a particular problem in real-life situations where drivers are reluctant to reduce their speed in
rural villages. In recent years this problem has been tackled using a variety of ‘village gateway
schemes’. Although these can be successful in the short run, the long-terms effects are less
convincing. The advisory system appears to have some success in reducing speed, although
this could be enhanced by providing the information about changes in speed limit before they
actually occur. This would provide drivers with the opportunity to decelerate before reaching
the lower speed limit.
Speed control also demonstrates other benefits in urban areas such as maintaining lower speeds
on curve negotiation and in areas where there are vulnerable road users. Speed variance was
also reduced under speed control, suggesting that widespread implementation could have the
effect of smoothing traffic flow by reducing extreme speed values.
It was particularly encouraging to find that the advisory system worked almost as well as the
dynamic speed limiter in potentially hazardous situations such as sub-standard curvature and
poor visibility conditions. Such benefits may be important considering the proportion of
accidents that occur in such conditions.
However, there were also some negative effects of the speed control systems. Firstly, in the
case of car following, it was found that those driving with a control system spent more time at
shorter headways. This may have been due to impatience, or due to drivers keeping their foot
on the accelerator to maintain maximum speed. Such driving behaviour may result in a higher
incidence of rear-end collisions, especially if drivers experience a degree of complacency.
The incidence of collisions at the end of the trials was found to be higher in the fixed condition,
independent of speed. This may reflect some degree of loss of situational awareness, such that
drivers experiencing the fixed speed limiter were taken out-of-loop, and faced with a critical
situation found it consequently more difficult to react in time. This is supported by the
relatively late braking that was observed in the fixed condition.
There were no effects of systems on reported mental workload. However it should be noted
that any reduction in workload may have been offset by the LCD display, which could have
caused additional processing. Future studies should involve the monitoring of eye movements
in order to ensure that distraction by the in-car display is not detrimental to the primary task of
driving.
25
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
The acceptability exercise demonstrated driver’s dislike for a system that potentially controls
their speed. However it was found that after experiencing such a system, drivers were less
negative towards it. As predicted, drivers are more inclined to find systems that do not exert
control as more acceptable, as they did the advisory system in this experiment.
In conclusion, speed control appears to offer benefits of mean speed and speed variance
reductions. In addition speed adaptation behaviour in rural villages could greatly benefit from
such a system from decreases in village entry speeds. However these benefits should be
considered in the light of potential safety costs such as reduced following distances and
possible complacency and loss of vigilance.
26
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
5. REFERENCES
Almqvist, S. and Nygård, M. (1997). Dynamisk hastighetsampassning - Demonstrationsförsök
med automatisk hastighetsreglering i tätort., Bulletin 154. Lund University.
Carsten, O.M.J., Groeger, J.A., Blana, E. and Jamson, A.H (1997). Driver performance in the
EPSRC Driving Simulator: A validation study. Final report to EPSRC Contract No.
GR/K56162.
Comte, S.L. (1996). Response to automatic speed control in urban areas: a simulator study.
Institute for Transport Studies, University of Leeds. ITS Working Paper, no. 477.
Comte, S.L. Wardman, M. and Whelan, G. (1997). Acceptability of External Vehicle Speed
Control. DETR project: External Vehicle Speed Control. Deliverable 5. Institute for Transport
Studies, University of Leeds.
Dahlstedt, S. (1994). The SARTRE-tables. Opinions about traffic and traffic safety of some
European drivers. VTI report 403. Linköping, Sweden.
Davidsson, F. (1995). Dynamisk hastighetsanpassning - scenario 2020. TFK. PM.
De Waard, D., Jessurun, M., Steyvers, R., Raggatt, P. and Brookhuis, K. (1995). Effect of
road layout and road environment on driving performance, drivers’ physiology and road
appreciation. Ergonomics 38(7) pp1395-1407.
Kulmala, R., and Beilinson, L., (1993). The effects of speed limiters of heavy vehicles - A
traffic simulation study. VTT, Technical Research Centre of Finland.
Ministry of Transport, Public Works and Water Management. Directorate Road Safety.
(1997). Intelligent Speed Adaptation. The structural control of driving speeds by means of
technological support for the drivers of motor vehicles.
Nilsson, L., (1996). Effekter av adaptiva farthållare, Sektion A. In: Gustafsson, P., (Ed.)
ARENA, Test site West Sweden. Learning and demonstrating new technology and design
principles for road transport, Göteborg, Sweden.
Perrett, K.E. and Stevens, A. (1996). Review of the potential benefits of Road Transport
Telematics). TRL Report 220. Transport Research Laboratory, Crowthorne, UK.
Persson, H., Towliat, M., Almqvist, S., Risser, R., and Magdeburg, M., (1993). Hastighetsbegränsare i bil. Fältstudie av hastigheter, beteenden, konflikter och förarkommentarer vid
körning i tätort. Department of Traffic Planning and Engineering, Lund, Sweden.
Richardson, J.H., Ward, N.J., Fairclough, S.H., & Graham, R. (1996). PROMETHEUS/
DRIVE AICC Safety Assessment: Basic Simulator). DoT Vehicle Standards and Engineering
Division Contract DPU 9/81/1 , HUSAT Research Institute, UK.
27
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
Saad, F. and Malaterre, G. (1982). La regulation de la vitesse: Analyse des aides au controle
de la vitesse. ONSER.
Várhelyi, A. (1996). Dynamic speed adaptation based on information technology: a theoretical
bacground. Bulletin 142. Department of Traffic Planning and Engineering, University of Lund,
Sweden.
Ward, N.J., Fairclough, S. and Humphreys, M. (1995). The effect of task automatisation in
the automotive context: A field study of an Adaptive Intelligent Cruise Control System. International Conference on Experimental Analysis and measurement of Situation Awareness,
Daytona Beach, Florida (Nov 1-3).
28
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
APPENDIX A - Experimental materials
(1)THE NASA-RTLX QUESTIONNAIRE
The initials TLX stand for Task Load indeX and this questionnaire is designed to assess your
own feelings and perceptions about the difficulty and mental workload associated with the
experimental task.
The questionnaire divides workload into a number of contributing factors and all these factors
add up to the total difficulty of the task. Please read the definitions of each factor carefully
before completing the questionnaire.
DEFINITION OF 6 FACTORS WHICH DESCRIBE THE LOADS PLACED ON AN
INDIVIDUAL DURING THE DRIVING TASK
MENTAL DEMAND
This refers to the ‘thinking’ component of the driving task. For example, consciously making
decisions about the traffic environment or deciding how to respond to the scenarios. How
much of this type of thinking, deciding, calculating, remembering, looking, searching, etc. did
you need to do? Was the task easy or demanding, simple or complex in this respect?
PHYSICAL DEMAND
How much physical activity was required (e.g. operating brake, clutch and accelerator, steering
the vehicle, using the indicator, etc.)? Was the task easy or demanding, slow or brisk, slack or
strenuous in this respect?
TIME PRESSURE
Did you feel you had enough time to adequately perform the experimental task?
PERFORMANCE
How satisfied were you with your performance in achieving the goals of the experimental task
i.e. safe driving?
EFFORT
How hard did you have to work (mentally and physically) to achieve your level of
performance? Did you feel stretched or comfortable during the task?
FRUSTRATION LEVEL
How insecure, discouraged, irritated, stressed and annoyed versus secure, gratified, content,
relaxed and complacent did you feel during the driving task?
A-1
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
Project : MASTER 3.2
Date: .......................
Subject ID ................
Condition :....………
RTLX MENTAL WORKLOAD
Please place a vertical line through each scale to indicate your level of workload on each
of the six factors.
Mental Demand
LOW 
HIGH
Physical Demand
LOW 
HIGH
Time Pressure
LOW 
HIGH
Performance
POOR 
GOOD
Effort
LOW 
HIGH
Frustration Level
LOW 
HIGH
A-2
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
(2)ACCEPTABILITY QUESTIONNAIRE
Project : MASTER 3.2
Date: .......................
Subject ID ................
Condition :
Pre-test
Imagine driving your car if it was fitted with a speed advisory system. This system would
display messages advising you of the appropriate speed for the particular area through which
you are driving.
Please indicate how acceptable you would find such a system by ticking a box on every line on
the scale below.
useful
pleasant
bad
nice
effective
irritating
assisting
undesirable
raising alertness
useless
unpleasant
good
annoying
superfluous
likeable
worthless
desirable
sleep-inducing
A-3
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
Project : MASTER 3.2
Date: .......................
Subject ID ................
Condition :
Post-test
We would like to know your opinion about the speed advisory system you have just used in the
simulator. Please indicate how acceptable you found the advisory system by ticking a box on
every line.
useful
pleasant
bad
nice
effective
irritating
assisting
undesirable
raising alertness
useless
unpleasant
good
annoying
superfluous
likeable
worthless
desirable
sleep-inducing
Was the information provided on the display:
very legible
very clear in meaning
very influential on your behaviour
completely illegible
completely unclear in meaning
had no effect on your behaviour
In your opinion, would this system make people drive more safely?
o yes, definitely
o yes, probably
o no, probably not
o no, definitely not
In your opinion, would this system make people commit less offences?
o yes, definitely
o yes, probably
o no, probably not
o no, definitely not
A-4
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
Would you have this system installed in your own car on a voluntary basis if it cost in the
region of £50?
o yes, definitely
o yes, probably
o no, probably not
o no, definitely not
Can you foresee any other possible advantages of such a system?
……………………………………………………………………………………
……………………………………………………………………………………
……………………………………………………………………………………
……………………………………………………………………………………
……………………………………………………………………………………
……………………………………………………………………………………
……………………………………………………………………………………
Can you foresee any possible drawbacks of such a system?
……………………………………………………………………………………
……………………………………………………………………………………
……………………………………………………………………………………
……………………………………………………………………………………
……………………………………………………………………………………
……………………………………………………………………………………
……………………………………………………………………………………
Thank-you for completing this questionnaire.
A-5
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
APPENDIX B - System operation
Using the logical road network, each individual section of road can be given a speed limit
which the simulator car will, if required, adhere to.
• If a change in maximum speed is required (for example travelling from a rural to an urban
zone) and the speed of the simulator car is in excess of the speed limit, it is decelerated by
the formula:
a=
vl − v
Tc
where
a is the drone’s acceleration (m/s/s)
vl
and
v
Tc
is the speed limit of a particular road section (m/s)
is the present speed of the drone (m/s)
is time constant of this first order system (1.5s)
until the new maximum speed is attained.
• If the simulator car is travelling at less than the speed limit then the speed limiter is inactive.
If the subject attempts to accelerate to above 30 mph the vehicle dynamics model automatically
prevents any further increase in speed by closing the throttle and applying a small brake
pressure to the hydraulic system of 10 bar. Thus even if the driver depresses the accelerator to
its full extent, there results in no increase in speed.
The in-car display, is shown below.
SPEED CONTROL
ACTIVE
B-1
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
APPENDIX C - Statistical analyses
Mean speed
Village 1 (48)
Village 2 (48)
Village 3 (48)
Single carriageway
(96)
Dual
carriageway(112)
Motorway (112)
S.D speed
Village 1 (48)
Village 2 (48)
Village 3 (48)
Single carriageway
(96)
Dual
carriageway(112)
Motorway (112)
Max. speed
Village 1 (48)
Village 2 (48)
Village 3 (48)
Single carriageway
(96)
Dual
carriageway(112)
Motorway (112)
Levene
test
F-test
F value
Sig.
of F
0.00
0.38
0.00
0.55
0.912
0.44
0.59
0.455
0.72
2.88
0.04
0.04
Levene
test
0.01
0.92
0.12
0.59
F-test
F value
Sig.
of F
5.72
25.77
1.50
Kruskal-Wallis test
Chi-square Sig. of
χ²
26.81
0.00
39.09
0.00
23.45
0.00
Kruskal-Wallis test
Chi-square Sig. of
χ²
37.51
0.00
0.00
0.00
0.22
0.01
10.96
0.17
0.00
31.31
0.00
Levene
test
0.00
0.00
0.00
0.55
F-test
F value
Sig.
of F
0.524
Kruskal-Wallis test
Chi-square Sig. of
χ²
42.87
0.00
21.58
0.00
40.97
0.00
0.67
0.00
18.23
0.00
0.01
25.09
0.00
C-1
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
APPENDIX D- Driver comments
Advisory system
An audible signal (a beep) would be good to alert of speed limit changes. The system may
distract from the road.
Its a good reminder if you miss the speed signs.
You could end up loosing your own judgement skills.
Gives you early warning of dangers which is good for planning.
The system serves as a ‘conscience’, to remind you of the speed limit, but might take your
attention from the road.
Once a driver becomes familiar with the road the system would be ignored.
This sort of system would be ignored by the same type of people who ignore speed limits.
It would be also useful if the driver could have feedback on whether they are violating the
speed limit.
It made me feel more obliged to drive at the speed limit than I normally do.
Fixed system
If all cars were fitted with the system it would alleviate pressure from other motorists to drive
faster.
The system would have to tell you about all possible hazards on the road as drivers would
become reliant on it.
Drivers may take fewer risks when overtaking once they learn that they cannot suddenly
accelerate. Might be fuel savings and environmental benefit.
Once people are used to driving more slowly, they would do so in general. Some objections
will occur when it is first introduced. Those used to driving over the speed limit would tend to
see maximum limits as minimum ones and therefore possibly driver dangerously in some
circumstances.
Drivers might become complacent.
May prevent some accidents, especially in built up areas.
Could free up police time to concentrate on other crime. Means you don’t have to worry about
breaking the speed limit. Can cause frustration.
D-1
Evaluation of in-car speed limiters:
Simulator study
May 1998
Samantha Comte,
University of Leeds, U.K.
System is good for persistent offenders, but dangerous when overtaking.
There should be less congestion with such a system, and cars would be cheaper. But people
will try and by-pass it.
Could be driver error through expecting the system to take complete control, i.e. changing
gear, steering etc.
Dynamic system
This sort of system might be good for newly qualified drivers.
You can driver closer to the car in front with this system, but you need to be able to accelerate
out of bends.
It was really frustrating to use and made me really aggressive towards other road users. I was
really mad when there were slow cars in front of me because I had wasted enough time
already. I think it made me feel safer than I actually was, especially in built up areas.
Technology is a wonderful thing, however, it costs and can go wrong.
In some cases it is useful to have full control of the car, i.e. more power to get out of an
accident situation.
Could curb speeding, but need a greater margin of speed potential, i.e. need to be able to
squeeze a few more mph out of the car in difficult situations. Extremely frustrating and might
push user to committing other offences, like jumping traffic lights.
Could reduce accidents in built up areas and seriousness of accidents. Might make people drive
up to the limit indicated when they might naturally driver slower.
Helpful to keep to the speed limit, but you need to be able to over-ride the system in case of
emergency situations, i.e. you have to accelerate out of a difficult situation in a limited speed
zone.
Could decrease the chances of joy-riders and other criminals from escaping from the police,
and act as a deterrent. Could decrease the chance of drink induced accidents. People will
compensate by becoming less alert and too dependent on the system, and irritated by not being
in control. People may take other risks to compensate for safe driving.
D-2