You are on page 1of 10

KSCE Journal of Civil Engineering (2014) 18(1):284-293 Transportation Engineering

Copyright ⓒ2014 Korean Society of Civil Engineers


DOI 10.1007/s12205-014-0363-z pISSN 1226-7988, eISSN 1976-3808
www.springer.com/12205
TECHNICAL NOTE

Descriptive and Parametric Analysis of Pedestrian Gap Acceptance


in Mixed Traffic Conditions
Satish Chandra*, Rajat Rastogi**, and Vivek R. Das***
Received July 18, 2012/Revised December 11, 2012/Accepted March 30, 2013

··································································································································································································································

Abstract

Crossing is an inevitable part of walking in which the desired path of pedestrians conflicts with vehicular traffic. These conflicts
can result in either delay or risk to the pedestrians. Due to insufficient designated crossing points on a highway or improper design,
most pedestrians are forced to cross at random unpredictable locations. As such, they create confusion and risks to themselves, as
well as to the drivers. Road accident statistics reveal that pedestrian injury and deaths are very high in India and improper gap
acceptance during crossing is a major contributing factor. Gap acceptance data collected at 17 locations in five cities of India are
analysed in this study to understand the variation in the gap acceptance behavior through descriptive and parametric analysis.
Descriptive analysis provided central tendency, dispersion and distribution of gap acceptance data, whereas, parametric analysis
resulted in identification of relationships between factors affecting accepted gap. Accepted gap was found decreasing with conflicting
traffic and crossing speed of the pedestrians. It also found varying with change in number of lanes on the road to be crossed. Critical
gap was estimated from accepted gap and crossing time distribution functions for varied traffic and spatial conditions.
Keywords: pedestrian, gap, mixed traffic, traffic lanes, India
··································································································································································································································

1. Introduction motorized fatalities in Mumbai were pedestrians (Mohan, 2004;


Grebert, 2008).
Pedestrians form an integral part of urban transportation system. Studies have revealed that there is a significantly lower potential
Walking had been and will remain as one of the most important for conflict if pedestrians cross at an intersection rather than
modes of transport. As a sustainable mode, it helps in controlling crossing at a midblock section (Zhenzong and Nambisan, 2009;
the air pollution caused due to high automobile usage. With the Kyriacod et al., 1999). But it was also found that the pedestrians
increase in heavier motor vehicles, traffic volumes and vehicle cross at a location convenient to them and where there was little
speeds on roads, undesirable problems have cropped up for the delay (Garder, 1989; Hamed, 2001; Holló et al., 1995; Sisiopiku
pedestrians. Most important among them is the impending risk due and Akin, 2003). Enforcements, restrictions and regulations were
to vehicular interaction while crossing a road or while walking on a found to have little effect on such ‘illegal’ behavior of the
carriageway or shoulder. During crossing or walking maneuver, pedestrians (Schonfeld and Musumeci, 2003; Kopelias et al.,
both the road users need to ascertain the available gap between the 2002). In such conditions, it becomes imperative to study the
two so as to ensure mutual safety. An incorrect decision by any of crossing behavior of the pedestrians which is vital in pedestrian-
the two road users will result in an accident. Globally, the vehicle interaction and increasing pedestrian risks. It should be
pedestrians are found to be more vulnerable in such conditions. studied in terms of gaps maintained between the two users to
Literature on this subject indicates that pedestrians accounted for minimize the risk. Such gaps are also influenced by the width of
nearly 65% of the total accidental deaths; 4378 pedestrian accidents the road to be crossed, the number of lanes to be crossed in one
were reported in the US in 2008 and accounted for 12% of traffic direction, the segregation of the traffic (directional or by vehicle
fatalities; pedestrian fatalities accounted for 14% of all traffic category), the traffic volume, and pedestrians characteristics and
fatalities in Europe; and the situation in developing countries is still their crossing behavior.
grim with pedestrian fatalities going as high as 84% (Kareem,
2003; FARS, 2008; ERSO, 2007; Sayer 1997; Mohan, 2004). The 2. Literature Overview
situation in India is no different. Higher rates were observed in
urban areas; 54% of all traffic fatalities in Delhi and 80% of all non- Literature is examined for the studies conducted on gaps

*Professor, Dept. of Civil Engineering, Indian Institute of Technology, Roorkee 247-667, India (Corresponding Author, E-mail: satisfce@iitr.ernet.in)
**Associate Professor, Dept. of Civil Engineering, Indian Institute of Technology, Roorkee 247-667, India (E-mail: rajatfce@iitr.ernet.in)
***Associate Professor, Dept. of Civil Engineering, MVJ College of Engineering, Bangalore 560-067, India (E-mail: vivekdurgadath@gmail.com)

− 284 −
Descriptive and Parametric Analysis of Pedestrian Gap Acceptance in Mixed Traffic Conditions

acceptance behaviour of the pedestrians, the effect of gender and simplified to three; single stage, two stages and rolling. In the
age on the accepted gaps, the effect of road width (or number of first case, the pedestrians cross the road, irrespective of its width,
lanes) to be crossed, the crossing pattern, and the traffic on the in one crossing maneuver. In the second case, they cross up to
road. These studies are discussed in the following paragraphs. median in one go and subsequently cross the far side. The
Moore (1953) found that the pedestrians who accepted shorter pedestrians keep searching for gaps between continuous flow of
gaps increased their crossing speeds. Those who accepted gaps vehicles by adjusting speed and direction of movement in the
of less than 3 s walked at a speed of 1.57 m/s while those third crossing pattern. The same approach is used in the present
accepting longer gaps of 7 s walked at 1.2 m/s. Cohen et al. study also to categorize the crossing patterns in India based on
(1955) found that 92% of pedestrians crossed a 7.0 m wide road observations taken under different traffic flow and road geometric
when the available gap was 7 s while no one crossed the road conditions.
when gaps were shorter than 1.5 s and everyone crossed the road Effect of road width is also studied by various researchers.
when gaps were 10.5 s or greater. DiPietro and King (1970) Wilson and Grayson (1980) analyzed the proportion of pedestrians
found that accepted gap becomes longer as the waiting time on accepting gaps of less than 2 s at a crossing with two-way traffic
the curb increases. Zhao and Wu (2003) analyzed microscopic flow. They found 3.4% of males and 2.1% of females accepting
behavior of pedestrians at signalised intersections, including these very short gaps when crossing against nearside traffic.
accepted gaps and lags in a mixed traffic situations. The average Palamarthy et al. (1994) found that pedestrians were more likely
acceptable gap and lag for the pedestrians were found to be 5.79 to look for an overall gap than separate gaps in individual traffic
s and 3.62 s, respectively. Das et al. (2005) found that critical streams. Oxley et al. (1997) reported that the average gap acceptance
gaps get influenced by the size of the intersection. Pedestrians for younger pedestrian was found to be 51.3 m and for older
were found to accept smaller gaps when crossing from the pedestrian this value was 69.1 m for two-way road. For one-way
median than from the sidewalk curb. Most of the pedestrians road, the average gap accepted by younger and older pedestrians
accepted a gap of 8 s. The summary of the gap accepted by the was 119.2 m and 134.1 m, respectively. Brewer et al. (2006)
pedestrians’ in different countries as reported by different researchers observed that pedestrians did not always wait to cross the street
is given in Table 1. The data indicate that the accepted gap varied when all lanes were completely clear; instead, they anticipated
between 1.5 s and 10 s. that the lanes would clear as they crossed and used a “rolling
Another aspect of interest is the influence of pedestrians’ gap” to cross the road. They found 85th percentile accepted gaps
characteristics on size of the gap accepted. Oxley et al. (1997) between 5.3 and 9.4s, with a trend of increasing gap length as
reported that some of the older pedestrians reject larger gaps than crossing distance increased.
their younger counterparts, but not longer enough thus making Andrew (1991) found that physical factors relating directly to
risky crossing decisions. Das et al. (2005) analyzed the revealed road-crossing behavior such as traffic volume, and conditions
preference data based on video recording of a crossing in India, relating indirectly to this behavior such as darkness and weather
and found that children and younger people accept gaps that conditions, are also likely to affect crossing behavior. Hine and
were rejected by the older persons, but found no difference in Russell (1993) observed that crossing strategies and pedestrian
gaps accepted by two genders. trip-making activity get modified in response to the changes in
The crossing patterns followed by the pedestrians are defined traffic conditions, namely traffic volumes and parking activity.
in many studies, but the patterns given by Song et al. (1993) look Carthy et al. (1995) found that older pedestrians had problems
wider in approach. They examined the interaction of pedestrians when crossing wide streets or intersections with dense traffic
and vehicles (gaps) at crossings which were at least 10 m away because it was difficult for them to look at traffic approaching
from a designated crossing. Their observations divided the from various directions. Oxley et al. (1997) also found that it was
pedestrians’ crossing tactics into four categories namely, “two- difficult for older pedestrians to assess the traffic on the farthest
gap”, “risk-taking”, “two-stage”, and “walk and look”. Brewer et carriageway.
al. (2006) observed that pedestrians’ crossing maneuvers can be The literature presented above highlights the need to study the

Table 1. Summary of Pedestrian Time Gaps Accepted in Different Countries


Researcher (Year) Country Accepted Gap (Seconds)
Moore (1953) United Kingdom 3.00 to 7.00
Cohen et al. (1955) United Kingdom 1.50 to 7.00
DiPietro and King (1970) United States of America 3.00 to 10.00
Palmarthy et al. (1994) United States of America 3.33 to 7.14
Oxley et al. (1997) Australia 51.30 m to 134.10 m
Zhao and Wu (2003) China 5.79
Das et al. (2005) India 8.00
Brewer et al. (2006) United States of America 5.30 to 9.40

Vol. 18, No. 1 / January 2014 − 285 −


Satish Chandra, Rajat Rastogi, and Vivek R. Das

pedestrian crossing behavior at designated or most favoured


midblock locations in developing countries like India. Also, the
influence of type of approaching vehicle and the impact of driver
behavior on gap acceptance is not adequately covered in the
literature.

3. Methodology and Data

The main emphasis of this study was on estimating the gaps


between a pedestrian crossing position at a midblock location
and the approaching vehicle, which are either accepted or
rejected by the pedestrian. In case the vehicle position is such
that a pedestrian decides and cross the distance safely then the
difference between pedestrian and vehicle position in space or
Fig. 3. Concept of Critical Gaps
time is referred as ‘gap accepted’. Otherwise, it is ‘gap
rejected’ and the pedestrian waits at crossing location for
another acceptable gap. This causes pedestrian delay. The This constituted the sample for gap analysis as well as for the
concept is defined in Fig. 1. distribution of gaps and crossing times. The crossing patterns of
The video camera was installed at the selected pedestrian the pedestrians’ were also observed under classified category as
crossing locations in different cities in such a way that it captures defined by Brewer et al. (2006) and shown in Fig. 2.
the pedestrians waiting on the curb as well as the approaching Cumulative distribution curves were plotted for gap
vehicles. The crossing movement of pedestrians and approaching acceptance (Fa) and crossing time (Ft) of pedestrians and the
vehicles were captured for an hour at each location. Later these intersection of the curves Fa and (1-Ft) provided the critical
data were played on a monitor using a digital clock to extract the gap as shown in Fig. 3.
information regarding pedestrians waiting time, rejected gaps The data were collected during 2008-2009 at 17 locations in
and accepted gaps. The approaching vehicle movement towards five cities of India, namely Delhi and Chandigarh in the North
the pedestrian location and pedestrians’ crossing maneuver were and Chennai, Erode and Coimbatore in the South. The weather
observed on the display unit. If the pedestrian does not cross and was fine and visibility was complete on the days of data collection.
keep waiting at the side of the carriageway then the time gap The cities were selected based on their population so as to
between first approaching vehicle and pedestrian location is represent all categories of city classes in India. The study locations
treated as gap rejected, whereas, if the pedestrian starts crossing within a city were chosen based on intensity of pedestrian
the carriageway, then the time gap between the nearest approaching movement, its midblock location, width of the road, number of
vehicle and pedestrian location is treated as the gap accepted. lanes in each direction, traffic volume and vehicle speeds. In the
case of divided road sections, the pedestrian crossing data were
collected on one side of the median and for an undivided road
section for the complete width of the carriageway. The details of
study locations are given in Table 2.
The classification of the road sections based on the number of
lanes and the direction of traffic resulted in the following:
a. Two Lane One Way (2L-1W): 13 test sections, with sample
size of 3689
b. Two-Lane Two Way (2L-2W): 2 test sections with sample
size of 986
Fig. 1. Gaps Accepted or Rejected c. Three-Lane Two Way (3L-2W): 2 test sections, with sample
size of 408
The total sample extracted for gap analysis was 5083. This
resulted in an average sample size of 299 per location with a
range of 41 to 652. Lower sample was at Marina beach in
Chennai and Rock Garden in Chandigarh. In both the cases, the
visitors were found reaching by vehicles and walking from
parking area to the site. This resulted in lower pedestrian crossing
movements. Exceptionally high pedestrian activity was witnessed
at Kaladipet, Chennai and Gandhipuram, Coimbatore, which are
Fig. 2. Definition of Crossing Patterns either shopping or mixed land use areas.

− 286 − KSCE Journal of Civil Engineering


Descriptive and Parametric Analysis of Pedestrian Gap Acceptance in Mixed Traffic Conditions

Table 2. Details of Pedestrian Crossing Locations


Width of Traffic Lanes Traffic Vehicle
Population Sample
City Location carriageway Land use and Directional Volume Approach
(millions) size
(meter) Movement (PCU/h) # Speed (km/h)
Connaught Place 8.4 Shopping 2-L 1-W1 1658 36.48 296
New Delhi 13.7 Kalkaji Mandir 10.4 Recreational 2-L 1-W2 2482 46.55 314
Near Andhra Bhawan 9.6 Residential 2-L 1-W2 2858 56.46 200
TH Road 11.6 Shopping 2-L 2-W1 2834 37.42 334
Old Washermanpet 7.25 Commercial 2-L 1-W2 1417 24.51 370
Chennai 4.34
Kaladipet 10.0 Shopping 2-L 2-W1 2510 19.14 652
Marina 7.0 Recreational 2-L 1-W2 1600 29.13 41
Gandhipuram 7.0 Commercial 2-L 1-W2 2395 24.06 512
Medical College 7.8 Mixed 2-L 1-W2 2079 24.84 258
Coimbatore 0.93
Near City Bus Stand 7.8 Mixed 2-L 1-W2 1663 25.35 206
100th ft Road 7.0 Mixed 2-L 1-W2 2203 23.06 262
Rock Garden 12.8 Recreational 3-L 2-W1 1634 41.86 120
Sukna Lake 13.6 Recreational 3-L 2-W1 1569 41.80 288
Chandigarh 0.90 Sector 17 10.6 Commercial 2-L 1-W2 892 36.52 470
Bus Station, ISBT 9.0 Mixed 2-L 1-W2 1586 33.51 350
Near PGIMS 9.5 Educational 2-L 1-W2 2191 38.83 244
Erode 0.15 Near CSI School 8.6 Educational 2-L 1-W2 2243 29.71 166
1
Undivided road section
2
Divided road section, considered only one side for analysis
ISBT: Inter State Bus Terminus
#
Traffic volume in one direction in the case of divided carriageway
PGIMS: Post Graduate Institute of Medical Sciences

Table 3. Accepted Gap (s) at Different Locations and Road Conditions


Standard Lower Upper
Operational Classification Mean Skewness Kurtosis Minimum Maximum
Deviation Percentile Percentile
Rock Garden 11.98 4.55 1.37 1.17 5.41 36.22 9.1 19.9
Sukna Lake 11.28 3.68 1.54 2.57 6.16 51.90 11.9 28.8
3L 2W 11.63 4.1 1.46 1.87 5.41 51.90 10.15 24.35
TH Road 7.30 6.21 3.0 12.86 2.03 47.57 3.8 13.6
Kaladipet 10.36 3.51 1.18 1.30 4.34 22.96 7.3 14
2L 2W 8.83 4.86 2.09 7.08 2.03 47.57 5.55 13.8
Sector 17 10.58 2.62 1.57 4.24 3.34 29.53 3.34 15.63
Bus Station, ISBT 10.13 2.91 1.65 3.72 3.52 20.05 5.5 10.9
Near PGIMS 8.78 2.87 0.97 1.28 3.99 18.85 5.9 11.7
Old Washermanpet 12.78 4.30 1.47 2.59 5.66 30.57 9.6 17.9
Marina 7.04 3.41 1.85 2.85 4.30 17.30 4.30 17.30
Gandhipuram 10.94 5.07 2.53 9.38 4.04 38.87 7.1 14.7
Medical College 7.15 2.96 1.31 2.01 1.91 18.63 4.5 9.8
Near City Bus Stand 9.05 3.49 1.59 4.11 3.24 23.93 6.3 12.1
100th ft Road 6.69 2.32 1.70 4.71 2.87 17.14 4.9 9.4
Connaught Place 8.95 4.68 2.43 7.41 3.27 32.53 5.7 12.1
Kalkaji Mandir 8.30 2.96 1.52 2.96 3.35 20.96 5.9 11.6
Near Andhra Bhawan 8.60 3.75 1.29 1.67 3.24 21.87 5.5 12.6
Near CSI School 6.38 3.81 2.92 10.50 3.00 25.85 3.8 8.5
2L 1W 8.87 3.47 1.75 4.42 1.91 38.87 5.56 12.63

4. Descriptive Analysis package and various descriptive parameters like mean, standard
deviation, skewness, etc. were estimated. The analysis was done
Descriptive analysis was carried out to estimate the central with respect to different pedestrian characteristics like age and
tendency, dispersion and distribution of the accepted gaps at gender, and type of crossing. Statistical significance at 95%
various locations. The analysis was carried out using SPSS confidence interval was tested between different parameters at

Vol. 18, No. 1 / January 2014 − 287 −


Satish Chandra, Rajat Rastogi, and Vivek R. Das

each location. The difference between the time gap accepted by two genders
reduced to less than 5 percent when carriageway changed from
4.1 Average Accepted Gap 2L1W to 2L2W. It however, increased to 13 percent on a 3L2W
Accepted gaps by pedestrians and their descriptive statistics carriageway. In the case of female pedestrians, these figures are
for all the locations and road conditions are given in Table 3. 4.5% and 29% respectively. It indicates that male pedestrians are
Average gap accepted in the case of 3-L 2-W road varies affected by both number of lanes and two-way traffic whereas,
marginally with an average of 11.63 s. Both of the locations in female pedestrians become safety conscious only when number
this case were recreational. An increase in the width of the road of lanes increases. These facts can be corroborated from the F-
and decrease in the traffic volume had resulted in acceptance of test statistics also given in Table 5.
relatively lower time gaps. In the case of 2-L 2-W road, the Further classification of pedestrians was done based on their
average gap is 24% lower as compared to the 3-L 2-W road. It age. The pedestrians were classified into 3 groups as young (less
was observed that relatively lower increase (13%) in traffic than 20 years), middle age (20 to 50 years), and old age (more
volume (above 2500 PCU/h) as compared to the width of the than 50 years) as perceived through video. The sample size was
carriageway (16%) had resulted in acceptance of substantially 352 young, 1095 middle age and 397 old age pedestrians for 2-L
lower gaps, the reduction being 30%. The analysis of two-lane 1-W carriageway system; 78 young, 301 middle age and 114 old
one-way (2-L 1-W) sections indicated an average gap of 8.87 s age pedestrians for 2-L 2-W carriageway system, and 519
with a wide range of 6.38 s to 12.78 s. The gap accepted was young, 1491 middle age and 531 old age pedestrians for 3-L 2-W
found to be higher in commercial land uses, a value of 10 s and carriageway system. The average time gaps accepted by the
above. Time gap values of less than 8 s were observed in areas pedestrians in these age groups are given in Table 6. In general,
with either educational or near recreational areas. Mixed land old age pedestrians were found accepting larger gaps as compared
uses had values in a range of 8 s and 10 s. to pedestrians in other age groups. This might be the effect of
All distributions were found skewed positively, thus indicating their age related crossing capabilities as well as their higher
that most pedestrians preferred a safety margin beyond the safety consciousness. However, F test indicated that the increases
average time gap. The maximum time gap accepted by the in time gaps for all the three age groups are not statistically
pedestrians was found to be more than 17 s. This happened al significant at 95% confidence level if only traffic became two
locations where traffic was low. Minimum time gap was found to directional on a 2-L carriageway. But with the change in number
be 1.91 s, indicating a very risky crossings maneuvers existing in of lanes, the variation in time gaps accepted by the three age
mixed traffic condition. groups is significant.

4.2 Gaps Accepted in Gender and Age Groups 4.3 Impact of Crossing Pattern
The time gap data were classified based on gender. The sample It was observed that the pedestrians cross the carriageway in
size was 1353 males and 492 females for 2-L 1-W carriageway three different manners. These were single stage, two stages and
system, 252 male and 241 female for 2-L 2-W carriageway and rolling, as defined earlier. The observed crossing maneuvers for
143 male and 61 female for 3-L 2-W carriageway system. The the three types of operational carriageways are given in Table 7.
male pedestrians accepted shorter gaps than female pedestrians The single stage crossings dominated on a 2-L carriageway and
on all types of roads as shown in Table 4. Similar results are rolling dominated on 3-L carriageway. It was interesting to note
reported by Cohen et al. (1955) also, but it was not discussed that on a 2-L 1-W carriageway substantial percent of pedestrians
with respect to the operational classification of the carriageway. were adopting two-stage crossing maneuver, which is not as per

Table 4. Average Gaps Accepted by Gender Groups


Male Female
Carriageway
type Standard Lower Upper Standard Lower Upper
Mean Mean
Deviation Percentile Percentile Deviation Percentile Percentile
3L 2W 10.25 3.41 5.41 51.90 11.6 3.67 9.82 35.64
2L 2W 8.63 5.28 2.14 47.57 9.00 4.43 2.03 32.98
2L 1W 8.54 3.46 1.91 38.87 9.08 3.95 2.62 37.06

Table 5. F-test Statistics for Gender Classified Pedestrians


Male Pedestrians Female Pedestrians
Comparison
Statistics Remarks Statistics Remarks
2-L 1-W v/s 2-L 2-W F(1, 1361) = 4.009, P<0.05 Significant F(1, 641) = 0.786, P>0.05 Not Significant
2-L 1-W v/s 3-L 2-W F(1, 1195) = 201.389, P<0.05 Significant F(1, 419) = 27.416, P<0.05 Significant
2-L 2-W v/s 3-L 2-W F(1, 326) = 75.650, P<0.05 Significant F(1, 244) = 31.530, P<0.05 Significant
Note: Statistics computed at 95% confidence level.

− 288 − KSCE Journal of Civil Engineering


Descriptive and Parametric Analysis of Pedestrian Gap Acceptance in Mixed Traffic Conditions

Table 6. Average Time Gaps Accepted by Pedestrians in Different Age Groups


Young Middle aged Old aged
Location
Mean SD LP UP Mean SD LP UP Mean SD LP UP
Rock Garden 10.5 3.71 5.41 29.86 11.6 4.44 7.12 36.22 15.3 5.47 - -
Sukna Lake 10.2 4.96 8.33 37.30 11.0 3.47 6.16 51.90 14 3.67 12.77 35.64
3L 2W 10.35 4.34 11.3 3.96 14.65 4.57
TH Road 6.66 4.23 2.69 27.98 7.37 7.04 2.03 47.57 10.06 7.23 2.39 22.33
Kaladipet 9.37 3.71 5.07 20.73 10.53 3.51 4.34 22.96 10.79 3.10 6.12 20.62
2L 2W 8.02 3.97 8.95 5.28 10.43 5.17
Sector 17 8.9 3.24 - - 10.52 5.24 3.34 29.53 10.68 4.38 5.00 15.78
Bus Station, ISBT 7.49 2.65 5.13 12.85 7.97 3.05 3.84 20.05 7.68 2.75 3.52 16.02
Near PGIMS 9.10 4.43 4.06 18.85 8.99 2.84 3.99 17.27 8.13 2.13 4.66 11.96
Old Washermanpet 11.54 4.36 5.66 30.57 12.91 4.23 7.72 27.78 14.41 4.22 7.45 24.17
Marina 6.92 4.40 3.84 17.30 6.04 1.77 4.30 11.14 7.86 2.94 4.99 11.40
Gandhi-puram 10.16 4.16 4.04 28.95 11.04 5.26 4.89 38.87 11.11 5.47 5.47 33.23
Medical College 5.91 2.70 3.00 12.47 7.15 3.02 1.91 18.63 7.39 2.97 2.73 14.26
Near City Bus Stand 9.27 3.02 2.63 15.86 8.91 3.92 3.24 23.93 8.47 3.06 2.86 15.64
100th ft Road 5.66 1.74 3.67 11.09 6.63 2.43 2.87 17.14 7.61 1.99 5.16 11.95
Connaught Place 10.12 5.48 5.63 18.32 9.14 5.41 3.55 32.53 8.21 2.77 2.68 15.72
Kalkaji Mandir 7.04 2.26 3.59 14.93 8.39 2.89 3.35 19.35 9.43 3.51 6.14 20.96
Near Andhra Bhawan 8.20 3.90 3.24 18.29 8.88 3.92 3.89 21.87 7.51 2.76 2.25 12.40
Near CSI School 5.52 3.51 3.53 17.54 6.83 4.56 3.00 25.85 6.32 3.05 3.00 17.49
2L 1W 8.14 3.53 8.72 3.73 8.83 3.23
SD: Standard Deviation; LP: Lower Percentile; UP: Upper Percentile

Table 7. Different Crossing Maneuvers on the Carriageways accepted during rolling and two stage maneuvers were higher
Carriageway Single stage Rolling Two Stage than those in single stage maneuver. It is quite obvious also as
operation (%) (%) (%) two stage and rolling crossings require more time to judge the
2 – lane one way 49 31 20 availability of the gap in the second lane. In the case of 2-L 1-W
2 – lane two way 65 33 02 carriageways, accepted gaps were only 6-7% higher for rolling
3 – lane two way 41 49 10 and two-stage crossing when compared with single stage crossings.
However, making the traffic two-way affected the rolling gaps
substantially, around 18 percent. This indicates that pedestrians
normal belief. Micro level examination of crossing maneuvers first evaluate the traffic in the near side lane and after that search
by location indicated that the two-stage crossing maneuver for a gap in rolling maneuver in the next traffic lane. The
highly dominated (52% to 73%) at locations where the carriageway statistical comparison of the gaps under different crossing
width varied between 7.0 and 8.6 m, and traffic volume ranged patterns is given in Table 8. As may be seen, making the traffic
between 1600 and 2250 PCU/h. But it could not be generalized two-way on a 2-L carriageway did not make any significant
as two locations falling in these ranges had negligible two-stage change in gaps accepted. The increase in number of lanes, on the
crossings. Under two-way traffic on a 2-L carriageway, the share other hand, has effect on accepted gaps.
of single stage crossings increased substantially. With further
increase in the number of lanes, the rolling became substantial. 5. Parametric Analysis
Further analysis of gaps data accepted by the pedestrians
during different types of crossings indicated that the gaps Parametric analysis was done to study the influence of various

Table 8. F-test Statistics for Gaps Accepted During Different Crossing Patterns
Single stage Rolling Two stage
Comparison
Statistics Remarks Statistics Remarks Statistics Remarks
2-L 1-W v/s 2-L 2-W F(1, 332) = 2.244, P>0.05 Not Significant F(1, 237) = 0.713, P>0.05 Not Significant F(1, 431) = 0.005, P>0.05 Not Significant
2-L 1-W v/s 3-L 2-W F(1, 286) = 67.723, P<0.05 Significant F(1, 998) = 121.423, P<0.05 Significant F(1, 330) = 40.557, P<0.05 Significant
2-L 2-W v/s 3-L 2-W F(1, 98) = 27.045, P<0.05 Significant F(1, 351) = 71.664, P<0.05 Significant F(1,113) = 28.893, P<0.05 Significant
Note: Statistics computed at 95% confidence level.
Parametric Analysis
Parametric analysis was done to study the influence of various factors on time gaps as accepted by the pedestrians under different operating conditions
on the carriageways.

Vol. 18, No. 1 / January 2014 − 289 −


Satish Chandra, Rajat Rastogi, and Vivek R. Das

factors on time gaps as accepted by the pedestrians under This defined the inverse function between the two. To eliminate
different operating conditions on the carriageways. the undefined form of the relationship if the conflicting flow rate
is zero or is too high (approaching infinite value) an additive
5.1 Effect of Conflicting Traffic constant was introduced in the relationship. This resulted in a
Microscopic analysis was carried out to study the interaction relationship as given by Eq. (2).
between pedestrians and vehicular traffic. Instead of estimating
B1
traffic flow rates on a carriageway based on 15 min or 1 hr data, AGap = B0 + -----------
- (2)
C Flow
the concept of instantaneous conflicting traffic flow as suggested
by Kyte et al. (1991) is used. This concept examines vehicle on a where,
minor road seeking a gap in traffic on the major road at a two AGap = Accepted gap, s
way stop controlled intersection. This concept is being extended CFlow = Conflicting flow rate, vps, and
to midblock crossing locations also where pedestrians waiting to B0 and B1 are the coefficients
cross the carriageway are considered synonymous to vehicle on a
The model parameters for three types of carriageway are given
minor road and vehicles plying on the carriageway define the
in Table 9. ANOVA indicated that the estimated parameters are
instantaneous conflicting flow. The flow rate is estimated by
statistically significant and of plausible sign. The models developed
considering the observation time as given by Eq. (1).
for three types of carriageways were found to be significantly
n different. The goodness of fit was low in the case of 2-L
Conflicting flow rate = ---------------- (1)
( tn – to ) carriageways but was satisfactory in the case of 3-L carriageway.
The typical scatter plots of the relationships for different carriageway
where,
operations are shown in Fig. 4.
n = Number of observed conflicting vehicles for the
subject pedestrian, including the conflicting vehicle
passing just after the pedestrian Table 9. Model Parameters for Conflicting Flow Rate Relationship
tn = Time of arrival of nth conflicting vehicle at reference Type of Model Parameters
point R2 F Value
carriageway Bo B1
to = Time of arrival of a pedestrian at the kerb 3L 2W 5.441 1.63 0.693 212.499
Intuitively, it is assumed that as the conflicting flow rate 2L 2W 4.485 1.356 0.337 169.274
increases the time gaps accepted by the pedestrians will reduce. 2L 1W 4.635 12.46 0.336 482.58

Fig. 4. Conflicting Flow Rate Versus Accepted Gap: (a) Two-lane One-Way Carriageway, (b) Two-lane Two-way Carraigeway, (c) Three-
Lane Two-Way carriageway, (d) Combined Data

− 290 − KSCE Journal of Civil Engineering


Descriptive and Parametric Analysis of Pedestrian Gap Acceptance in Mixed Traffic Conditions

Table 10. Model Parameters for Crossing Speed Relationship 6.1 Average Critical Gaps
Carriageway Model Parameters 2 Average critical gap at different locations were estimated and
R F-value
type bo b1 are given in Table 11. The critical time gaps in the case of 2-L 1-
3L 2W 17.885 0.924 0.541 175.574 W carriageways were found varying between 3.44 and 9.75 s.
2L 2W 11.159 0.770 0.630 629.459 The variation in critical gaps on different sections is attributed to
2L 1W 9.396 0.792 0.554 2406.945 traffic volume and/or land use of adjoining area. Traffic volume
on these sections ranged between 1600 and 2900 PCU/h in
educational, residential or recreational areas and it resulted in
5.2 Effect of Crossing Speed lower critical gaps of 3 to 6s. It increased to 6 - 8 s in the mixed
It was assumed that the pedestrians would accept shorter gaps or shopping areas with traffic volume ranging between 1600 and
only when they are in a position to cross at a faster speed. This 2400 PCU/h, and further to 9 s and above when the traffic flow
indicates towards an inverse relationship. The effect of speed was 1400 to 2200 PCU/h in purely mixed land use areas. As may
may or may not be considered as linear and therefore a power be observed traffic volume on these sections do not vary much
function would be more appropriate. The functional form and hence, the main influencing factor is land use. On 2-L2W
between time gap and crossing speed is given by Eq. (3). carriageways, the critical gap reduced by around 21% with
b1 almost no change in average time gaps. This is the result of
AGap = b0 x ( VC ) (3)
increased friction of two-way traffic. Further increase in the
Where, number of traffic lanes (3L-2W) increased the critical gaps by
AGap = Accepted gap (s) 118% which is quite high. All these locations are in the recreational
Vc = Crossing speed (m/s) land use with traffic volume in the range of 1500-1700 PCU/h.
b0 and b1 are the coefficients
6.2 Effect of Gender and Age
The model parameters and corresponding statistics for different
Summary of critical gaps for pedestrians on three types of
locations are given in Table 10. The relationships for all the
carriageway as categorized by their gender and age are given in
carriageway types were found to be significant at 95% confidence
Table 12. As reported in literature also, the female pedestrians
level.
were found accepting higher gaps than male pedestrians. This
might be due to their lower crossing speeds and tendency of
6. Critical Gaps

Critical gap at each site was estimated from their gap


Table 12. Critical Gap (s) for Classified Pedestrians on Different
acceptance (Fa) and crossing time (Ft) distribution curves as Carriageways
demonstrated in Fig. 3. The intersection of the curves (Fa) and
Type of Middle
(1-Ft) gives the value of the gap which is just equal to the Male Female Young Old
Carriageway Aged
crossing time of a pedestrian. This is the critical gap. The 3-L 2-W 10.4 11.85 10.25 11.3 12.9
analysis was carried out for variation in critical gap with
2-L 2-W 5.05 5.35 7 5.3 5.3
respect to pedestrian characteristics, road width and traffic
2-L 1-W 6.54 6.78 5.95 6.68 7.38
volume.

Table 11. Average Critical Gaps on Different Carriageways


Location Critical Gap, s Average Gap, s Location Critical Gap, s Average Gap, s
Rock Garden 11.86 11.98 Sector 17 8.84 10.58
Sukna Lake 10.82 11.28 Bus Station, ISBT 9.40 10.13
3-L 2-W 11.34 11.63 Near PGIMS 9.75 8.78
Old Washermanpet 9.38 12.78
TH Road 3.08 7.33 Marina 4.00 7.04
Kaladipet 7.34 10.36 Gandhipuram 7.14 10.94
2-L 2-W 5.21 8.85 Medical College 4.7 7.15
Near City Bus Stand 6.13 9.05
100th ft Road 4.78 6.69
Connaught Place 6.87 8.95
Kalkaji Mandir 6.18 8.30
Near Andhra Bhawan 5.58 8.60
Near CSI School 3.44 6.38
2-L 1-W 6.63 8.87

Vol. 18, No. 1 / January 2014 − 291 −


Satish Chandra, Rajat Rastogi, and Vivek R. Das

maintaining larger safety margins than males. The movement of


traffic in both the directions on 2-L carriageways reduced the
time gaps by 21-23 percent for both genders. It is attributed to
increased vehicular friction. With the increase in number of
lanes, the time gaps accepted by pedestrians, in both the gender
classes, increased by almost two times ad it was further higher in
the case of female pedestrians.
The study of time gaps accepted by pedestrians in different age
groups indicated an increase in time gaps with the increase in the
age of the pedestrians. This increase was of the order of 10 to
12% in the case of 2-L 1-W carriageways and 10 to 14% on 3-L Fig. 6. Effect with Respect Traffic Volume
2-W carriageways. Increase in number of traffic lanes also
caused an increase in time gaps in the range of 10 to 14% within
respective age groups. Interestingly, on 2L2W carriageway, the traffic stream at higher volumes which force the pedestrians to
critical gap is same for middle age and old pedestrians while it is accept smaller gaps.
higher for old pedestrians on a 2L1W road or 3L2W roads. The relationship between the critical time gaps and traffic volume
is given by Eq. (5).
6.3 Relation with Width of the Carriageway
11576.95
The variation in critical gaps with carriageway width is shown CGap = 3.025 + ----------------------
Q (5)
in Fig. 5. As expected, the critical time gaps increases with 2
increase in the width of the carriageway. This is obviously due to ( R = 0.393, F ( 1, 14) = 9.046, p = 0.009 )
combined effect of number of traffic lanes to be crossed, traffic where,
volume plying in a section and the crossing speed of the CGap = Critical Gap, in seconds
pedestrians under the influence of different frictions. Q = Traffic volume, PCU/h
A mathematical relation between critical gap and width of the
carriageways is given in Eq. (4). 7. Conclusions
1.186
C Gap = 0.525xW
(4) A detailed analysis was carried out in this paper to determine
2
( R = 0.628, F ( 1.15 ) = 27.976, p = 0 ) various parameters that influence pedestrian crossing behaviour.
Various design values of gaps are presented through descriptive
where,
analysis and the relation between the influencing parameters are
CGap = Critical Gap, in seconds
developed. The study was carried out on three categories of
W = Width of the carriageway, in meters
carriageways, namely 2-L 1-W, 2-L 2-W and 3-L 2-W. The
study demonstrates the effect of carriageway width (number of
6.4 Relation with Traffic Volume
lanes), directional movements, traffic volume and pedestrians’
The variation in the critical gap with traffic volume is shown in
characteristics on the gaps accepted by the pedestrians’ as well as
Fig. 6. As may be seen, critical gap decreases with increase in the
on the critical gaps, which have design implications.
traffic volume. This decrease is quite rapid in the traffic volume
Analysis of gap data on three types of carriageway indicated
rage of 1000-3000 PCU/h and thereafter gets stabilizes. The
more effect of carriageway than traffic volume on gap accepted
decrease in critical gap is due to smaller gaps available in the
by a pedestrian. Accepted gap showed almost no change when
carriageway width reduced by 15 percent and traffic volume
increased by 78 percent. However, accepted gaps at the same
traffic volume level increased by 21 percent when carriageway
width was reduced by 32 percent. The effect of land use was also
analysed and the accepted gap was high in commercial land use
and low (less than 8 s) in educational and recreational land uses.
The effect of inherent capabilities of pedestrians was visible in
the results obtained from the analysis of gap acceptance data
based on gender and age. At macro level, the results of this study
are in line with those reported by Cohen et al. (1955), DiPietro
and King (1970), Oxley et al. (1997) and Das et al. (2005).
However, significant changes were observed during micro level
analysis. Female pedestrians were more safety conscious on
Fig. 5. Variation in Critical Gaps with respect to Carriageway Width wider roads than males. The critical gap for middle and old aged

− 292 − KSCE Journal of Civil Engineering


Descriptive and Parametric Analysis of Pedestrian Gap Acceptance in Mixed Traffic Conditions

pedestrians are same on 2L2W carriageway while it is higher for crossings.” Safety Science, Vol. 38, No. 1, pp. 63-82.
old aged pedestrians on 2L1W roads. Traffic conditions (one- Hine, J. and Russel, J. (1993). “Traffic barriers and pedestrian crossing
way or two-way) also influenced the pattern of crossing. With behaviour.” Journal of Transport Geography, Vol. 1, pp. 4, pp. 230-
239.
one-way movement, more number of pedestrians looked for
Hollo, P., Papp, I., and Siska, T. (1995). “Observation of elderly pedestrians
two-stage gaps, whereas single stage crossing was prevalent on on signalized crossings and of jaywalkers in the vicinity of pedestrian
two-way roads. The effect of increase in carriageway width was subways.” Proceedings of the 8th Workshop, International Cooperation
a shift to rolling pattern. The change in traffic operation affected on Theories and Concepts in Traffic Safety, Paris, pp. 1-11.
rolling gaps more as compared to other patterns. This indicates Kareem, A. (2003). “Review of global menace of road accidents with
that the pedestrians in India are more alert and continuously look special reference to Malaysia - A social perspective.” Malaysian
for a possible opportunity to cross the road. Journal of Medical Science, Vol. 10, No. 2, pp. 31-39.
The critical gap of a pedestrian was also influenced by traffic Kopelias, P., Papaioannou, P., and Vasiliadou, J. (2002). “Effectiveness
of a pedestrian measure in an arterial street in Thessaloniki, Greece:
volume, number of lanes and directional movement of the traffic.
First results.” Proceedings of the 13th Workshop, International
In general, critical gap on a 2-L 1-W was 20 percent lower than Cooperation on Theories and Concepts in Traffic Safety, Corfu, pp.
the average gap accepted. It was 41 percent lower on 2L2W 183-193.
system. However, accepted gap was almost same as the critical Kyriacod, M. C., Sun, W., Chien, S. I., Eisdorfer A., and Qu, T. (1999).
gap on 3-L 2-W carriageways. This might be the indication of “Effect of midblock access points on traffic accidents on state highways
relatively more homogenous behavior of pedestrians under in new jersey.” Transportation Research Record 1665, National
increased restrained crossing conditions. Mathematical equations Research Council, Washington, DC, pp. 75-83.
are suggested relating critical gap with average traffic volume on Kyte, M., Clemow, C., and Khisty, C. J. (1991). “Capacity and delay
characteristics of two way stop controlled intersection.” Transportation
the road. The analysis presented in this paper clearly brings out
Research Record 1332, National Research Council, Washington,
the variations in crossing behavior of pedestrians under diversified D.C, pp. 160-167.
effects. The results will be useful to traffic and transportation Mohan, D. (2004). The road ahead: Traffic injuries and fatalities in
planners and engineers. india, Transportation Research and Injury Prevention Programme,
IIT Delhi, India.
References Moore, R. L. (1953). “Pedestrian choice and judgment.” Journal of the
Operational Research Society, Vol. 4, pp. 3-10.
Andrew, H. W. (1991). “Factors influencing pedestrian cautiousness in Oxley, J., Brian Fildes, Elfriede Ihsen, Judith Charlton and Ross Day,
crossing streets.” Journal of Social Psychology, Vol. 131, No. 3, pp. (1997). “Differences in traffic judgments between young and old
367-372. adult pedestrians.” Accident Analysis and Prevention, Vol. 129, No.
Brewer, A. M., Fitzpatrick, K., Whitacre, A. J., and Lord, D. (2006). 6, pp. 839-847.
“Exploration of pedestrian Gap-acceptance behaviour at selected Palamarthy, S., Mahmassan, H., and Machemehl, R. (1994). “Models of
locations.” Transportation Research Record 1982, National Research Pedestrian Crossing Behavior at Signalized Intersections.” Research
Council, Washington, D.C, pp. 132-140. Report 1296-1. Center for Transportation Research, University of
Carthy, T., Packham, D., Salter, D., and Silcock D. (1995). Risk and Texas Austin.
safety on the roads: The older pedestrian, Report Prepared for AA Sayer, I. A. and Palmer C. J. (1997), “Pedestrian accidents and road
Foundation for Road Safety Research, University of Newcastle safety education in selected developing countries.” 3rd African Road
Upon Tyne, UK. Safety Congress, 14-17 April, Pretoria, South Africa.
Cohen, J., Dearnaley, E. J., and Hansel, C. E. M. (1955). “The risk taken Schonfeld, C. and Musumeci, A. (2003). Pedestrian travel: Getting
in crossing a road.” Journal of the Operational Research Society, queenslanders walking safely, Centre for Accident Research and
Vol. 6, No. 2, pp. 120-128. Road Safety-Queensland, Brisbane.
Das, S., Mansk, C. F., and Manuszak, M. D. (2005). “Walk or wait? An Sisiopiku, V. P. and Akin, V. (2003), “Pedestrian behaviors at and
empirical analysis of street crossing decisions.” Journal of Applied perceptions towards various pedestrian facilities: An examination
Econometrics, Vol. 20, pp. 529-548. based on observation and survey data.” Transportation Research
DiPietro, C. M. and King, L. E. (1970). “Pedestrian gap-acceptance.” Part F, Vol. 6, No. 4, pp. 249-274.
Highway Research Record, No. 308, NCHRP, Washington, D.C., Song, L., Dunne, M. C., and Black, J. A. (1993). “Models of delay and
pp. 80-91. accident risk to pedestrians.” Transportation and Traffic Flow Theory,
European Road Safety Observatory (ERSO) (2007). Traffic safety basic Proceedings of 12th International Symposium, Elsevier Science.
facts data base, Project Data base of the European Commission, Wilson, D. G. and Grayson, G. B. (1980). Age-related differences in the
Director- General Transport & Energy. road crossing behaviour of adult pedestrians, Report No. LR 933,
Fatality Analysis Reporting System (FARS) (2008). Data summary, Transport Research Laboratory, UK.
Report by US Department of Transportation. Zhao, Jiali and Jiaping, Wu (2003). “Analysis of pedestrian behavior
Garder, P. (1989). “Pedestrian safety at traffic signals: A study carried with mixed traffic flow at intersection.” Intelligent Transportation
out with the help of a traffic conflicts technique.” Accident Analysis Systems Journal, IEEE, Vol. 1, pp. 323-327.
and Prevention, Vol. 21, No. 5, pp. 435-444. Zhenzhong, C. and Nambisan, S. S. (2009) “Methodology for evaluating
Grebert, J. (2008). Pedestrian safety consideration enhancement, SICA the safety of midblock pedestrian crossing.” Transportation Research
Project Proposal Report, SIMBA Project. Record 1828, National Research Council, Washington, D.C., pp. 75-
Hamed, M. M. (2001). “Analysis of pedestrians’ behavior at pedestrian 82.

Vol. 18, No. 1 / January 2014 − 293 −

You might also like