REVIEW OF THE RED TO GREEN SEQUENCE AT TRAFFIC

advertisement
REVIEW OF THE RED TO GREEN SEQUENCE AT TRAFFIC SIGNALS
Alastair Maxwell
TRL
Iain York
TRL
1. SUMMARY
At traffic signals in the UK a red-with-amber signal (‘starting amber’) denotes
the impending change to green, but conveys the same prohibition as the red
signal. It is fixed at two seconds. Practice varies across Europe with some
countries using a starting amber and some not. The period for which the
starting amber is shown also varies by country, most use either one or two
seconds. Additionally some European countries have or are considering
reducing the length of the starting amber.
The UK Department for Transport commissioned TRL to review omitting or
reducing the ‘starting amber’ both in efficiency and safety terms, compared
with the current policy.
The research included: an international literature review; previous accident
studies at traffic signals; a review of practice in other countries; stakeholder
interviews and questionnaires; video analysis of vehicle, cyclist, and
pedestrian response to the current starting amber, and conflict analysis; and a
TRL Driving Simulator experiment examining driver responses to 0, 1 and 2
second starting amber periods.
The advantages and disadvantages of a change to the current starting amber
policy are assessed, including effects on non-motorised traffic.
2. BACKGROUND AND OBJECTIVES
While the starting amber is a requirement for the UK, it is not the case in
many countries abroad. Drivers are becoming increasingly familiar with traffic
signal systems and strategies used elsewhere.
The last significant UK research on the starting amber was carried out in the
early 1960’s. The Department for Transport finds that it is increasingly difficult
to defend the status quo based upon this, given changes in vehicle and signal
technology and driver experiences. As such this new research has been
commissioned to investigate the consequences of omitting or amending the
length of the starting amber, both in safety and efficiency terms.
Note that, any views expressed are not necessarily those of the Department
for Transport.
© Association for European Transport and contributors 2005
3. METHODOLOGY
The main strands of the research were:
•
An initial review including practice in other countries, an international
literature review, previous accident studies at traffic signals, stakeholder
interviews and questionnaires, and an initial assessment of the potential
implications on road users (including pedestrians).
•
Comprehensive video analysis of the current UK situation, recording
vehicle and cyclist movements in relation to the starting amber, conflict
analysis, and investigating how pedestrians use the starting amber.
•
A pedestrian questionnaire, used to ascertain their perceived crossing
behaviour at traffic signals and their views on a change to the starting
amber.
•
A TRL Driving Simulator experiment examining driver’s responses to 0, 1
and 2 second starting amber periods. Drivers’ decisions were investigated
when they were approaching, or already stationary at the junction. At
some of the junctions the drivers could just see the signal shown by the
opposing signal heads. Drivers completed questionnaires before and after
the trials.
Further methodology details are given in the relevant sections.
4. STARTING AMBER POLICY
Pan-European policy on the use of the starting amber is laid down in the
European rules concerning road traffic signs and signals (European
Conference of Ministers of Transport, 1974), commonly referred to as the
‘Vienna convention’. The starting amber is optional and no timing periods are
specified. The red and amber shown together means ‘that signal is about to
change, but shall not affect the prohibition of passing indicated by the red
light.’
The use of the starting amber varies across Europe, generally the ‘low
countries’ and southern European countries do not use a starting amber,
while Scandinavian and northern European countries do, as shown in Table 1.
The period for which the starting amber is shown also varies by country from
between one and three seconds.
© Association for European Transport and contributors 2005
Table 1: European starting amber practice
Starting amber
Length of starting
used
amber
(seconds)
Poland
1
Finland
1 to 1.5 (2 before
1990)
Sweden
1.5 (1 in the future)
Greece
Germany
1
Luxembourg
UK
2
Netherlands
Austria
2
Portugal
Czech Republic
2
Italy
Cyprus
2
Spain
Iceland
2
Ireland
Malta
2
Russia
3
Starting amber not used
Belgium
France
Both the Finnish and Swedish decisions to reduce the starting amber were
made subjectively on the basis of drivers violating the starting amber. An
interviewed Finnish traffic control committee member stated that there has
been no noticeable effect on delay, and both drivers and signal experts are
satisfied with the practice used now.
Most countries outside Europe do not utilise the starting amber, including:
USA; Canada; Australia and New Zealand.
Until around 1980, two Australian States, Western Australia and Victoria, had
starting ambers. They were removed in line with the national standards. An
interviewed representative of Victoria State Highway Authority was of the
opinion that drivers easily adapt to some signals having a starting amber and
some not.
5. PREVIOUS STUDIES ON THE EFFECTS OF THE STARTING AMBER
5.1 Introduction
There have been three reported on-street trials where the starting amber
period was omitted. The UK trials previously had a three second starting
amber, the others a two second period:
•
•
UK trials analysing 18 junctions from 1959 to 1962 (Older, 1963)
Hong Kong trials covering 2 junctions and 2 pedestrian crossings in 1973
(Seneviratne, 1974).
© Association for European Transport and contributors 2005
•
Melbourne, Australia, trials covering 7 junctions in 1965; analysis was only
undertaken at two junctions (McGill, 1970).
There was also study in London in 1979 (Branston, 1979), at 10 approaches
to signalised junctions, which assessed the effect of the length of the starting
amber upon green starting lost time.
5.2 Effect on efficiency
The starting amber provides a warning of the impending green to drivers, and
this helps maximise the capacity of the junction by reducing starting lost time
(the time taken from the onset of green for queued vehicles to reach
maximum flow rate over the stopline (‘saturation flow’)).
The trials indicated average increases in vehicular starting delay of between 1
and 1.7* seconds (*3 second starting amber) when the starting amber is
omitted.
The increase in starting delay was less when opposing signal heads could be
seen. Older stated that if all the signals had been effectively screened, the
average effect of omitting the starting amber would have been greater, and
McGill (1970) found an initial 1.4 second increase in starting lost time at one
study junction which fell to zero after a period of weeks when drivers adapted
their behaviour to anticipate the start of green from opposing signal heads.
Older calculated that the increase in lost time would reduce junction capacity
by 6%, based upon the junctions used in this experiment. After the Hong
Kong trials the starting amber was maintained, the effect on vehicle delay was
cited as the main reason.
Branston (1979) maintained the starting amber, but studied the effect of its
length. The results indicated that the increase in starting lost time is
approximately equal to the reduction in starting amber time, when the starting
amber period is reduced.
5.3 Effect upon traffic signal violations
The starting amber has the same prohibition as the red signal. However,
previous surveys and work undertaken in this project indicate that there is a
notable violation rate.
The UK trials (Older, 1963) showed a reduction from 1.5 percent of all
vehicles entering the junction in the starting amber period to 0.1 percent in the
equivalent three seconds when the starting amber was omitted. However,
there was an increase in the number of vehicles entering the junction in the
stopping amber period, but not the following red. Prior to the omission 15
percent of the signal cycles had at least one vehicle entering the junction after
the amber appeared, while after the omission this figure had risen to 20
percent (a 33% increase). Older believed that this was due to drivers
© Association for European Transport and contributors 2005
becoming familiar with, and taking advantage of, the increase in start-up delay
of opposing vehicles.
The Hong Kong trials (Seneviratne, 1974) showed a reduction from 22.1% to
6.5% for first vehicle waiting at the lights to ‘jump the signals’. While this is a
significant reduction, it still shows a notable proportion of drivers who were
prepared to proceed, anticipating the onset of green, when there was not a
starting amber.
The Branston (1979) results indicate that the level of starting amber violations
increase with the length of the starting amber period.
5.4 Effect upon safety
The UK and Hong Kong trials indicated potential safety benefits from omitting
the starting amber from the increase in the effective intergreen time caused by
the increase in starting delay. Seneviratne noted a significant reduction in
pedestrian conflicts with vehicles at pedestrian crossings (which had low
clearance times and very high pedestrian flows), and the UK trials showed a
significant reduction in personal-injury accidents between vehicles from
different roads (47%), however this was offset by an unexplained increase in
single vehicle and same direction vehicle accidents (31% increase).
However, these trials had intergreen times which were well below the current
recommended values, and as such these benefits would be unlikely to be
shown in today’s conditions. Newby (1961) showed an 83% reduction in
accidents between vehicles on different roads by adding a one or two second
all-red period.
Increasing the effective intergreen potentially provides more time for
pedestrians to cross. However, pedestrians no longer receive a warning of the
start of vehicle green from the traffic signal indications. Branston found that
pedestrians were more likely to be crossing at the start of green when the
traffic signal indication was impossible or difficult to see, and the UK trials
found a 128% (Older 1960) increase in the number of pedestrians crossing in
the period that had previously been red-with-amber – part of this increase will
have also been due to a reduction in vehicles beginning to move in this period.
The UK trials showed that overall effect on pedestrian accidents was
negligible. Both Older and Seneviratne expressed concerns that there may be
an increase due to the lack of warning. Older suggested the negligible effect
on pedestrian accidents could have been due to the ‘change in sequence
made pedestrians rely more on personal observation of vehicle behaviour for
their safety than on the implication of the vehicle signal faces’.
5.5 Amber arrows
Amber arrows have been trialled in the UK to attempt to reduce the number of
false starts caused by drivers reacting to the wrong red-with-amber signal at
double-headed signal arrangements using green arrows. The trials indicate a
© Association for European Transport and contributors 2005
reduction in drivers responding to the wrong signal, but an increase in
violations of the stopping amber and following red.
It was thought that the lower visibility of the amber arrow was the likely reason
for the deterioration of driver behaviour at the end of green. The amber arrow
is less easy to see than a full roundel, making it more difficult to distinguish
and thus increasing driver response time, and reducing general compliance to
the signal (Pleydell and Gillam, 2002). A possible solution under trial is an
‘intelligent’ LED amber arrow, which shows an arrow for the starting amber
but a full roundel for the stopping amber.
A potential advantage of omitting the starting amber would be to prevent
drivers reacting to the wrong amber roundel under green arrow arrangements,
and thus negate the perceived need for amber arrows.
6. STAKEHOLDER INPUT
The main aim of the stakeholder input was to help identify the potential issues
arising from any change to the starting amber. A secondary aim was to get a
sample of stakeholders’ views on the subject.
TRL specialists covering traffic signals, cycling, motorcycling and pedestrians
were interviewed, and a list of potential issues associated with the starting
amber and its potential amendment drawn up.
The external stakeholder input covered the committee of the Traffic Control
User Group (TCUG) and a selection of Local Authorities. These stakeholders
were sent a questionnaire to complete and interviewed.
The stakeholders were in favour of maintaining the current two-second
starting amber (a minority were unsure whether two seconds was optimal)
and did not believe that a change would be worthwhile or the cost justifiable.
The main concerns, with regards to changing the current policy were:
•
•
•
•
•
The increase in starting delay should the starting amber be omitted
The effect upon co-ordination of closely spaced signals e.g. signalised
roundabouts – the starting amber signals the impending green, facilitating
the progression of approaching drivers.
The lack of warning to pedestrians of the onset of vehicle green.
Increased pressure on drivers to get away quickly at the start of green.
Implementation costs of any change
The only issue with the starting amber raised was that some drivers treat it as
‘go’. However, the effect of this was not regarded as being so significant as to
be a safety issue, and no interviewees were aware of any cases where the
intergreen time had been set above the guideline values, given in TA/16/81
General principles of control by traffic signals, due to drivers beginning to
proceed in the starting amber period.
© Association for European Transport and contributors 2005
7. RESULTS OF THE VIDEO SURVEYS SHOWING CURRENT RESPONSE
TO THE STARTING AMBER
7.1 Introduction
Videos of five junctions and two pedestrian crossings (Puffins) were analysed,
covering a range of site characteristics, including urban and interurban, and
low and high pedestrian flows.
The data were recorded for the first vehicle waiting at the stopline. Times
were recorded to 0.04 seconds (25 frames per second). The centre of the
front wheel was used as the position reference for the vehicle. Distances were
estimated to 0.5 metres (or less where possible) from junction plan/ measured
site dimensions. A clear Perspex sheet with sample distances from the
stopline was used on the screen to ensure consistency in reported distances.
7.2 Driver response to the starting amber
The results exclude situations where there was a pedestrian or vehicle conflict,
or blocking back within the junction i.e. where the starting movement could
have been impeded. Conflicts are analysed separately in Section 7.4.
The results also exclude twelve blatant red runners who did not wait for the
end of the opposing phase. Most of these were cyclists (7), or vehicles at the
junction ‘Camden’ (4). Camden has a double headed signal arrangement, with
the studied approach being a full signalised right turn green arrow, on some
occasions vehicles would proceed when the straight-ahead signal went green
when the right turn was still red. This type of violation is discussed in Section
5.5.
As can be seen from Table 2, a notable percentage of vehicles crossed the
stopline before the onset of green (36%), especially at junctions. On average,
vehicles are in front of the stopline by the time that the green starts, they start
to move 0.25 seconds before the onset of green and those that have stopped
behind the stopline cross it 0.38 seconds after green.
Table 2: Summary of driver response to the starting amber
© Association for European Transport and contributors 2005
Traffic signal
type
Analysis Type
Waiting
distance
(metres)
Start to
move time
(secs from
start of redamber)
Distance to
stopline at
the start of
green
(metres)
Distance
travelled by
the start of
green
(metres)
Cross
stopline
time* (secs Sample size
from start of
red-amber)
Waiting
behind
stopline
Cross
stopline
before
green*
Start to
move before
green
86%
36%
66%
83%
39%
63%
98%
27%
74%
Mean
0.68
1.75
-0.47
1.15
2.38
947
15th (85th) %ile
0.00
1.04
-2.50
2.50
1.44
2.35
0.36
1.81
-0.77
Mean
1.13
755
junctions
15th (85th) %ile
-0.50
1.04
-3.00
2.50
1.40
2.51
1.96
1.50
0.73
Mean
1.23
192
Puffins
15th (85th) %ile
0.00
0.88
-1.00
2.50
1.65
*Excludes vehicles that wait in front of stopline
Bold and underlined: Junction and Puffin results statistically different at the 95% level
Distance traveled is the 85th percentile, the rest are 15th percentiles
All
Comparing Puffins and junctions, the results indicate significant differences in
driver behaviour at the studied traffic signal installations. At Puffins, drivers
cross the stopline later and have a greater distance to the stopline at the start
of green. This is primarily due to the larger waiting distance. Possible reasons
for this are:
•
•
•
•
Drivers avoid blocking the pedestrian crossing.
Drivers avoid pressurising pedestrians.
There is not an Advanced Stop Line (ASL) to abuse as there was at one of
the junctions.
And perhaps the closely associated secondary signal heads means that
drivers get a better view from a distance back from the stopline.
The quicker start to move times at Puffins is likely to be due to:
•
•
A driver, seeing that the pedestrian has crossed, is likely to expect the end
of red shortly, also on seeing the pedestrian has crossed a proportion of
drivers (4%) then starting to move (See Table 3).
The relatively short red times at Puffins may mean that it is more likely that
the handbrake is not applied and the vehicle is still in gear.
Table 3 shows the percentage of drivers starting to move and crossing the
stopline in the first 0, 1, 1.5, and 2 seconds of the starting amber.
Table 3: Start to move and cross stopline times relative to time in starting
amber
© Association for European Transport and contributors 2005
Traffic signal Total sample
type
size
Waiting
behind
stopline
Cross
stopline
during red*
Cross
stopline
before 1sec
starting
amber*
Cross
stopline
before
1.5sec
starting
amber*
Cross
stopline
before
green*
Start to
Start to
Start to
move before move before
Start to
move during
1sec
move before
1.5sec
red
starting
green
starting
amber
amber
All
947
86%
0%
4%
17%
36%
1%
14%
40%
66%
junctions
755
83%
0%
4%
19%
39%
1%
14%
40%
63%
Puffins
192
98%
1%
2%
10%
27%
4%
18%
42%
74%
*Excludes vehicles that wait in front of stopline
Bold and underlined: Junction and Puffin results statistically different at the 95% level
Distance traveled is the 85th percentile, the rest are 15th percentiles
As can be seen, for all traffic signal types, 4% of vehicles that waited on or
behind the stopline crossed it before the first second of red-amber, and 17%
crossed it before 1.5 seconds of red-amber.
Figures 1 to 4 illustrate the times that vehicles start to move and cross the
stopline. Figure 5 shows the distances travelled by the start of green.
Figure 1: Start to move time
30%
26.50%
25.03%
Percentage of vehicles
25%
20%
17.85%
15%
10.67%
10%
7.60%
5.28%
5%
2.22%
1.90%
0.11% 0.11% 0.21% 0.32% 0.21% 0.21%
0.95% 0.53% 0.21% 0.11%
0%
-5.5
-4
-3
-2
-1.5
-1
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
5.5
Start to move time seconds from start of red-amber
© Association for European Transport and contributors 2005
Figure 2: Cumulative frequency graph of start to move time
100%
95%
Percentage of vehicles that have started to move
90%
85%
80%
75%
70%
65%
60%
55%
50%
45%
40%
35%
30%
25%
20%
15%
10%
5%
0%
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4 4.2 4.4 4.6 4.8 5 5.2 5.4 5.6 5.8 6
Seconds from start of red-amber
Figure 3: Cross the stopline time (excludes those stopped in advance of the
stopline)
30%
25.95%
Percentage of vehicles
25%
20%
18.97%
14.93%
15%
13.10%
10.53%
10%
6.36%
5%
3.30%
3.06%
0.12%
1.71%
0.73%
0.24%
0.61%
0.24%
0.12%
5
5.5
6
6.5
0%
-1
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
Crossing the stopline time seconds from start of red-amber
© Association for European Transport and contributors 2005
Figure 4: Cumulative frequency graph of time of crossing the stopline
(excludes those stopped in advance of the stopline)
100%
95%
90%
85%
80%
Percentage crossed the stopline
75%
70%
65%
60%
55%
50%
45%
40%
35%
30%
25%
20%
15%
10%
5%
0%
6.00
5.80
5.60
5.40
5.20
5.00
4.80
4.60
4.40
4.20
4.00
3.80
3.60
3.40
3.20
3.00
2.80
2.60
2.40
2.20
2.00
1.80
1.60
1.40
1.20
1.00
0.80
0.60
0.40
0.20
0.00
Seconds from start of red amber
Figure 5: Distance travelled by the start of green
40%
36.22%
35%
Percentage of vehicles
30%
24.29%
25%
21.01%
20%
15%
11.40%
10%
4.65%
5%
2.01%
0.32%
0.11%
4>5
5>6
6>7
0%
0
0>1
1>2
2>3
3>4
Distance travelled by start of green (metres)
Table 4 gives a summary of the results by vehicle type. The mean and
percentage results for ‘Cars and Taxis’ have been statistically assessed
against the other vehicle types.
© Association for European Transport and contributors 2005
Table 4: Summary of results by vehicle type
Vehicle type
Analysis Type
Waiting
distance
(metres)
Start to
move time
(secs from
start of redamber)
Distance to
stopline at
the start of
green
(metres)
Distance
travelled by
start of
green
(metres)
Cross
stopline
Total
time* (secs
sample size
from start of
red-amber)
Mean
0.68
1.75
-0.47
1.15
2.38
15th (85th) %ile
0.00
1.04
-2.50
2.50
1.44
Cars & Mean
0.93
1.75
-0.17
1.10
2.38
taxis
15th (85th) %ile
0.00
1.04
-2.00
2.50
1.48
0.68
Mean
1.94
-0.25
0.93
2.38
LGV
15th (85th) %ile
0.00
1.06
-2.00
2.15
1.41
Mean
1.55
1.55
0.71
2.72
0.84
HGV
15th (85th) %ile
0.00
1.04
0.00
1.95
1.92
0.63
3.32
Mean
1.28
1.99
0.66
PSV
15th (85th) %ile
-0.75
1.64
-1.75
1.00
2.15
-1.51
1.40
-3.91
2.41
1.59
Mean
Motorbike
15th (85th) %ile
-4.00
0.73
-6.00
4.00
0.87
Mean
-4.42
1.67
-5.69
1.27
1.56
Bicycle
15th (85th) %ile
-6.40
0.94
-8.00
2.00
1.56
*Excludes vehicles that wait in front of the stopline
Bold: Other vehicle types statistically different at the 95% level
Distance traveled is the 85th percentile, the rest are 15th percentiles
LGV: Light Goods Vehicle. HGV Heavy Goods Vehicle. PSV Public Service Vehicle
All
Waiting
behind
stopline
Cross
stopline
before
green*
Start to
move before
green
947
86%
36%
66%
727
91%
35%
65%
111
93%
38%
57%
18
100%
22%
56%
16
81%
8%
50%
62
31%
74%
87%
13
8%
100%
77%
The results show that motorcyclists and bicyclists have particularly different
characteristics from cars and taxis. The majority of both tend to wait in-front of
the stopline, and are quite far advanced beyond the stopline by the start of
green. Only one cyclist waited behind the stopline (this includes a junction
with an ASL), and not included in this analysis are seven blatant red running
cyclists.
The potential for rapid acceleration of motorcycles, together with infringing the
starting amber means that there is potential for conflicts in the junction unless
the motorcyclists check carefully before accelerating into the junction. The
conflict analysis indicates that this could be the case, with only one
motorcyclist being involved in a conflict and that was due to congestion within
the junction, there was no starting amber violation.
The results were also assessed by time of day (peak, off-peak and lunch),
generally there was little difference, except in the peak period slightly fewer
drivers wait behind the stopline and vehicles tended to be slightly further in
advance of the stopline at the start of green (0.70 metres).
Results by site were assessed. Site plans and signal timings were collected,
and site visits undertaken, particular reference was given to intervisibility (the
ability for drivers at each stopline to see each other and pedestrian waiting
positions) and the ability to see opposing signal heads. The results showed a
fair amount of variation in driver behaviour by site. An assessment of the
results was made against the site characteristics. The results would appear to
indicate that drivers modify their behaviour based upon the ability to assess
the risks and the potential for conflicts. Situations where intervisibility is good
and the potential for conflicts is low are treated with less caution. Busy urban
junctions had notably lower violation rates than open, low pedestrian flow,
© Association for European Transport and contributors 2005
interurban junctions. Ahead movements also had a quicker start to move time
than turning movements.
Previous research indicated that the ability to see opposing signal heads is a
significant factor affecting the starting delay of vehicles. The sites studied here
tended to provide little opportunity for drivers to see the opposing signals and
their visibility did not appear to be a particularly significant factor in this study.
7.3 Pedestrians
Pedestrian arrivals and crossing decisions at the crossing facilities were
recorded between the last two seconds of red and the first two seconds of
green. A full day (08:00 to 18:00) was analysed at each site.
The results showed that the vast majority of pedestrians were prepared to
cross the road in-front of stationary traffic, 80% of all pedestrians arriving
between the last two seconds red and first two seconds green chose to cross
in this period if the opposing vehicle was stationary. While if the pedestrian
arrives when vehicles have started to move then they are much less likely to
cross; only 30% in the starting amber and 10% during the green.
There was no statistically significant difference between the proportion of
pedestrians crossing at the end of red (89%) and the starting amber (84%).
Thus unlike the previous UK trials (Older, 1963) there is no indication from
these results that the starting amber signal acts as a warning to pedestrians of
the onset of green.
7.4 Conflicts
A conflict is defined as a situation where two road users would collide if
neither took any avoiding action, e.g. braked, swerved etc. The ‘conflict’
analysis recorded incidents occurring around the time of the starting amber in
an attempt to assess whether there are safety issues associated with driver
and pedestrian behaviour under a two second starting amber.
Two Puffins (one a single direction approach and the other with a two-way
approach) and five approaches to junctions were studied, with over 52 hours
of video analysed. In total 82 conflicts were recorded, 33 being vehicle-vehicle
conflicts and 49 vehicle-pedestrian conflicts; a conflict rate per hour of around
1.6 per approach/ Puffin around the start of green.
Table 5 shows the types of conflict. 90% of vehicles delayed start of
movement when there was a pedestrian conflict, and with a vehicle conflict
the percentage was about two-thirds; with the conflict point further away
vehicles were prepared to start moving and then brake if the conflict point was
still not clear. There were no near misses recorded
© Association for European Transport and contributors 2005
Table 5: Type of conflict
Type of conflict
Pedestrian
Vehicle
All conflicts
conflicts
conflicts
Count
%
Count
%
Count
%
Obstruction – delayed start of
44
90%
22
67%
66
80%
movement
Precautionary braking or lane
5
10%
11
33%
16
20%
change when risk is minimal
Near miss
0
0%
0
0%
0
0%
Very near miss
0
0%
0
0%
0
0%
Collision
0
0%
0
0%
0
0%
Total
49
100%
33
100%
82
100%
The majority of pedestrian-vehicle conflicts recorded were associated with
pedestrians beginning to cross in the starting amber period, as shown in
Table 6. Many pedestrians who started to cross towards the end of red were
able to clear the conflict point before a vehicle could have reached it. An
assessment was made of the studied sites and it was found that on average
only pedestrians crossing in the last second of red (half the length of the
starting amber period) would be in potential conflict with a starting vehicle.
Table 6: Period that conflicting pedestrian starts to cross
Pedestrian start Percentage
to cross period
Red
25%
Starting amber
60%
Green
15%
Total
100%
About half of pedestrians in a potential conflict increased their walking speed.
The most common combined action to avoid the conflict is for the pedestrian
to proceed normally and the driver to delay starting.
Only three pedestrian conflicts involved a starting amber violation, but in all
cases the conflict probably would have occurred without the violation, in one
the pedestrian starting crossing in green; in another the pedestrian ran,
allowing the violation without a safety impact; and in the third the vehicle
violated but proceeded very slowly.
Vehicle-vehicle conflicts occur when a vehicle has not cleared the conflict
point by the time vehicles from the next phase would have reached it. The
causes of not clearing the conflict point found in this study were:
•
•
•
•
Opposed right turners
Blocking back/ congestion
Red running
Large vehicles making a difficult turning movement
© Association for European Transport and contributors 2005
By far the most common event was for the starting vehicle to delay movement
and the conflicting vehicle to proceed normally.
Though not classified as near misses, there were four more potentially serious
conflicts. All involved at least partial obscuration between the starting and
conflicting vehicle due to blocking back through the junction, three were
associated with apparent red runners and one an opposed right turner. Three
of the four had starting amber violations.
In total five vehicle conflicts involved a starting amber violation (15%, which is
much lower than the 36% recorded for all starting vehicles), three of these
probably would not have occurred without the violation. Two involved red
running vehicles on other approaches and one an opposed right turn
movement combined with congestion within the junction.
Thus in conclusion where there is a visible potential conflict drivers tend not to
violate the starting amber, and if they do they can use precautionary braking
later, and, as shown in Section 7.3, pedestrians tend not to start crossing if
vehicle movement has started. The potentially more serious conflicts were
associated with the conflict being at least partially obscured from view and the
conflicting vehicle movement occurring well into the green of the new stage.
Of the total of 82 conflicts, 8 involved a starting amber violation and three
conflicts (3.7%) probably would have been avoided without the violation. In no
cases was a starting amber violation the sole or main cause of a conflict.
8. PEDESTRIAN QUESTIONNAIRE
The absence of a warning of the onset of green to pedestrians was seen as a
concern by interviewed stakeholders. Pedestrian behaviour was surveyed onstreet (Sections 7.3 and 7.4) and a pedestrian questionnaire devised to
attempt to assess anticipated changes in behaviour. A total of 107 participants
completed the questionnaire.
In general pedestrians believed they would feel less safe if the starting amber
was removed (73% at junctions with red/green men, and 81% at junctions
without). The main reasons given were: not receiving any warning of the
impending green; the starting amber gives time for drivers to check for
pedestrians; and a feeling that there would be less time to cross due to the
immediate change to green.
For the small percentage of participants who thought that they would feel
safer without a red-with-amber, one of the main underlying reasons was that
they would be more cautious and consequently safer. Another reason was the
inability of drivers to start off on red-with-amber.
Only 6% of pedestrians stated that they always wait for the green man. When
crossing against a red man 76% stated that they always or ‘usually’ check the
signal indication given to vehicles before deciding to cross. Most participants
© Association for European Transport and contributors 2005
(70%) considered that they would be more hesitant about crossing against a
red man if the starting amber was omitted. At junctions without pedestrian
phases 68% said that they would be less likely to start to cross part way
through the red to traffic.
In summary pedestrians consider that they would feel less safe if the red-withamber was removed and would show more caution and request pedestrian
phases more often. This could possibly cause greater delays to pedestrians
and traffic, but could potentially offset any negative influences on pedestrian
safety, as indicated by the Older (1963) trials.
9. THE TRL DRIVING SIMULATOR EXPERIMENT
TRL’s driving simulator consists of a full-size car linked to a wrap around
projected screen. Motion is simulated through the suspension units of the
vehicle and provides heave, pitch and roll to the car body. It was used to
investigate drivers’ responses to changing the starting amber period in a safe
environment.
The simulated road network consisted of 15 signal controlled junctions, one
kilometre apart on a 50 mph speed limit dual carriageway.
There were automated cars travelling in the opposite direction and from the
side roads. No automated cars were on the same side of the road as the
driver. This prevented them influencing the drivers decision when to start
moving forward.
The traffic signals were set to change from red either on the approach to the
signals (20, 30 or 150 metres from the stopline) or after a varied period of
waiting at the stopline. At the signals where the drivers came to a stop, the
opposing signal head indication could just be seen at half the junctions. The
intergreen timings were set to six seconds at all junctions.
A total of 60 drivers, 30 male and 30 female took part in the simulator
experiment. Twenty participants of each gender were in the age range 25 to
55, whilst ten were older drivers.
The younger drivers took part in three separate driving runs, with the order
randomised. In each drive the starting amber duration was constant at all
junctions and set to either:
•
•
•
2 seconds (currently in use within the UK)
1 second
0 seconds (change straight from red to green)
The older drivers were only asked to drive twice, with the first and last of
these red-with-amber durations (i.e. the extreme situations). This examined
whether age affected reactions to red-with-amber duration, in addition to the
main sample examining the reactions in detail.
© Association for European Transport and contributors 2005
The results showed that drivers waiting at the stopline were delayed by:
•
•
•
1.4 to 1.6 seconds when the starting amber was omitted (younger drivers)
1.2 to 1.3 seconds when the starting amber was omitted (older drivers)
0.7 to 0.8 seconds when the starting amber was reduced to one second
(younger drivers)
This equates to a loss of junction capacity, all else being equal. For instance a
three stage junction, with a cycle time of 80 seconds would have a reduced
capacity of around 6% if the starting amber was omitted and about 3% if it
was reduced to one second, based upon the results of the younger drivers.
Given the congested nature of many UK junctions in the peak hours this could
have a significant effect upon vehicle delay, especially for an omission, unless
the intergreen timings are lowered in line with the increases in starting delays.
The difference in ages appears to be largely due to older drivers showing
more caution under current conditions. They were less likely to start to move
in the red-with-amber than the younger drivers.
The crossing the stopline results were similar for vehicles approaching the
stopline when the signals changed, and indicate that the starting amber can
be beneficial to maintain progression at closely spaced signals. For signal
changes at 20 and 30 metres from the stopline younger drivers took 0.8 to 0.9
seconds longer to enter the junction when the red-with-amber duration was
reduced to one second. When the red-with-amber was removed younger
drivers entered the junction 1.6 to 1.7 seconds later after the start of green,
and older drivers 1.2 to 1.3 seconds later.
The ability to view opposing signal heads was found to be a significant
determinant of the delay in starting from rest both in previous trials and in this.
Younger drivers’ average delay was reduced by 0.7 seconds and older
drivers’ average by 0.4 seconds when the opposing signals were visible. This
reduction in delay did not depend significantly on whether or not a starting
amber was used. Therefore, unlike previous trials (Older and McGill), the
simulator results did not show drivers using the opposing signals more when
the starting amber is omitted than when it is present. However, these were
short term results and further behaviour modifications could occur with
familiarity, as shown by McGill (1970) described in Section 5.2.
For vehicles waiting at the stopline acceleration used was found to be slightly,
but statistically significantly (at the 95% confidence level), greater when the
starting amber was omitted. The maximum acceleration used was on average
between 0.08 and 0.09 m/s2 greater, and the maximum acceleration was
reached 0.7 to 0.8 metres earlier. There is an implication that drivers
accelerated slightly harder when the starting amber was omitted, possibly as
an over-reaction to an unexpected change. However, these slightly higher
accelerations had little effect on speed once through the junction, with the
average difference in speed only being 0.3 to 0.4 km/h greater 50 metres after
the stopline.
© Association for European Transport and contributors 2005
The results for average brake release time (hand or foot brake) and
accelerator on times shown in the graphs below, show that drivers react
quicker to a change from red to green than from red to starting amber and
more uniformly (as shown by the standard deviation).
© Association for European Transport and contributors 2005
Figure 6: Average time brake released after end of red
Time (seconds)
Time brake released after signal change
3.5
3
2.5
2
1.5
1
0.5
0
0 seconds
1 seconds
2 seconds
0
20
30
150
Distance from stop line (m)
Figure 7: Average time accelerator pressed after end of red
Time accelerator pressed after signal change
Time (seconds)
4
Starting
amber length
3
0 seconds
2
1 seconds
2 seconds
1
0
0
20
30
150
Distance from stop line (m)
Figure 8: Standard deviation of brake release time after end of red
Time (seconds)
Standard deviation of time brake released after signal
change
Starting
amber length
1.4
1.2
1
0.8
0.6
0.4
0.2
0
0 seconds
1 seconds
2 seconds
0
20
30
150
Distance from stop line (m)
© Association for European Transport and contributors 2005
Figure 9: Standard deviation of accelerator pressed time after end of red
Time (seconds)
Standard deviation of time accelerator pressed after
signal change
Starting
amber length
2.5
2
1.5
1
0.5
0
0 seconds
1 seconds
2 seconds
0
20
30
150
Distance from stop line (m)
The level of signal violations was found to be very low in the simulator
experiment. On investigation it was found that on average drivers waited two
metres further back than on-street. It is believed that the close to perspective
and visibility of the stopline in the simulator partly caused this, and this
accounts for much of the difference in crossing time, but not all, reaction times
also appear to be slightly slower. It is possible that drivers on the road, with
other vehicles waiting behind them, feel more pressure to start moving as
soon as possible.
Drivers completed questionnaires before and after the simulation runs. Over
three quarters identified the correct definition of the starting amber, however
20% of drivers gave definitions which implied that one could proceed with
caution. Also one third of drivers did not believe it was bad practice to cross
the stopline during the starting amber.
Participants generally preferred the current two seconds of red-with-amber
time: 69% of those aged under 55 and 55% of those aged over 55.
Participants’ reasons were due to familiarity, and feeling more comfortable
owing to having more time to prepare to go. 18% of the younger driver
preferred a 1 second red-with-amber as they felt most comfortable with it and
it reduced hesitancy and reduced waiting time. Approximately a sixth of the
participants preferred removing the red-with-amber completely. It was mainly
the older driver who chose this option (30% of them). Their reasons were that
it was easier to understand and made them feel safer. However, overall
drivers felt more hurried in their decisions without a starting amber, which
could have safety implications.
© Association for European Transport and contributors 2005
10. IMPLEMENTATION ISSUES
In March 2000 there were an estimated 12,300 sets of traffic signal controlled
junctions and 13,800 isolated traffic signal controlled pedestrian crossings
(DTLR, 2001) – note that many of the crossings are ‘Pelicans’, which use a
flashing amber rather than a red-with-amber. Additionally, there are portable
traffic signal controllers and temporary traffic lights.
The implementation costs of any change in the starting amber period would
be significant, particularly as new firmware would be required in all existing
controllers, as the starting amber period is fixed. The reduction in starting
amber in Finland was implemented over 5 years. Most controllers had a
configurable starting amber period, so it was relatively easy to modify. The
intergreen timings were maintained i.e. the omitted red-with-amber was
replaced by red.
11. CONCLUSIONS
The starting amber provides a warning of the impending green, giving drivers
time to prepare to move. This helps maximise the capacity of the traffic
signals and reduces pressure to get away quickly at the start of green. It also
serves to maintain progression at closely linked signals.
The simulator trial showed a more uniform response to the start of green
when the starting amber was omitted, and some drivers said that such a
sequence was easier to understand.
With a two second amber, part of the efficiency benefits are from some drivers
crossing the stopline in the starting amber. While this is a concern and against
the regulations, the current recommended UK intergreen timings are generally
regarded as appropriate, and the safety assessment showed that drivers
usually delay movement when there is a potential conflict, and only tend to
proceed on the starting amber when it is safe to do so.
To maintain the current efficiency levels, if the starting amber was omitted, the
intergreen timings would need to be adjusted accordingly with the increase in
start up delay. This has potential safety concerns, especially for vulnerable
road users, as drivers would now be starting to proceed in the green period
which previously was starting amber, which has been shown to be treated
with caution. Reactions to green are quicker and drivers felt under more
pressure.
Additionally, the potential increase in drivers anticipating the end of red from
opposing signals and traffic movements, is unlikely to warrant a wholesale
reduction in intergreen timings. Increased anticipation also has safety
concerns. Such behaviour could lead to drivers concentrating on the side
roads as opposed to scanning the whole conflict zone, and could increase
false starts, particularly at signals with variable stage orders.
© Association for European Transport and contributors 2005
A reduction in the starting amber period would reduce the violation rate of
drivers crossing the stopline before the start of green, while still providing
some of the efficiency benefits. Current driver behaviour indicates that a
length of 1.5 seconds would reduce the violation rate to about 17% and a one
second starting amber would virtually negate violations. However, violations
are likely to slightly increase with the predicted quicker response time, and the
implementation costs of any change in the UK would be significant.
REFERENCES
Branston, D. (1979) Some factors affecting the capacity of signalised
intersections. Traffic Engineering and Control (1979-08/09) 360-6.
Department for Transport, Local Government and the Regions. (2001)
Transport Statistics Great Britain 2001. London: The Stationery Office.
European Conference of Ministers of Transport. (1974) European Rules
Concerning Road Traffic Signs and Signals. Paris: European Conference of
Ministers of Transport (Vienna 1968 – Geneva 1971-73).
McGill, W. A., (1970) Optimising the use of amber in traffic signals.
Proceedings of 5th ARRB Conference. 5(3), pp 95-110.
Newby, R.F., (1961) The effect on accident frequency at signal controlled
crossroads with an all-red period. Traffic Engineering and Control. 1961,
3(2) 102-103.
Older, S. J. (1960) The effect of the omission of the red amber indication at
traffic signals (Leicester). Research Note RN/3702/SJO. Crowthorne: Road
Research Laboratory (Unpublished).
Older, S. J. (1963) Omission of the red/amber period at traffic signals. Traffic
Engineering and Control (November 1963) 414-417.
Pleydell, M., and Gillam, J. (2002). Yellow arrows with and without intelligence.
Traffic Engineering and Control (2002, 10) 354-5.
Seneviratne, J. R. V. (1974) Omission of the starting amber at traffic light
signals – a before and after study. Technical Report 157, Public Works
Department, Traffic and Transport Survey Division, Hong Kong.
© Association for European Transport and contributors 2005
Download