PEDESTRIAN SAFETY AT UNSIGNALISED CROSSINGS
M1 Traffic and Transportation Engineering
DEPARTMENT OF CIVIL ENGINEERING
COLLEGE OF ENGINEERING
A pedestrian is a person travelling on foot, whether walking or running. Pedestrians are the vulnerable road users. Pedestrian vehicular interaction decides the vehicular flow characteristics especially where there is a large pedestrian crossing exhibit. Therefore the pedestrian behaviour and driver’s attitude to pedestrians are important in the safety point of view.
With so many vehicles on the road, pedestrians need to be aware of their surroundings and the “rules of the road” to avoid being hit by a car or other motor vehicles. Older pedestrians are at especially high risk. Night time is a particularly dangerous time because visibility is reduced, and both pedestrians and drivers are likely to be drunken. It has been observed that the pedestrians have great difficulty in crossing as most of the drivers don’t care for the waiting pedestrians in unsignalised crossing. The risk of older pedestrians being struck by a car or truck is particularly high at marked crosswalks with no traffic signal or stop sign. This might occur because marked crosswalks may give older pedestrians a false sense of security.
Usually traffic stream facilities are designed by considering the number of vehicles that may be present in the future. The capacity of the traffic stream is explained on the basis of total number of vehicles that can be present in the traffic stream. But in real the capacity of traffic stream is not only decided by the number of vehicles but also by the number of pedestrians. The pedestrian vehicular interaction is an area of traffic engineering where thorough study is required. The interaction not only leads to the delay or reduction in capacity but also causes accidents. The most unwanted effect of the interaction is accident. Since the pedestrians do not like delay, he may try to cross the road by taking the risk. This leads to the accidents as the driver is also not willing to be delayed. In the case of unsignalised crossings accidents are more as the drivers may not willing to give the right of way to the pedestrians.
The spectacular increase in the number of motor vehicles on the road has created a major social problem-the loss of lives through road accidents. The appalling human misery and the serious economic loss caused by road accidents demand the attention of the society and call for the solution of the problem. The traffic engineer is concerned because many features of the highway affect the safety of the pedestrians and that of the vehicle. There has been an increased emphasis on improving pedestrian safety. The desire to improve pedestrian safety extends to areas typically seen as being non-pedestrian friendly, such as the higher speeds and wider roadways. With changing traffic conditions due to increase in traffic volume and congestion, pedestrians' ability to cross the road safely may be affected adversely. Recent developments in geometric design features, traffic control devices, and technologies may improve pedestrian safety and access by addressing specific problems associated with roadway crossings. Although numerous treatments exist at unsignalised crossings, there is growing concern about their effectiveness. Thus, there is a need to identify and study selected treatments to determine their effectiveness.
1.2 . OBJECTIVES OF THE STUDY
The objectives of this study are to review
• the studies on pedestrian behaviour at unsignalised crossings
• the studies on pedestrian vehicular interaction at unsignalised crossings
• the passive pedestrian detection systems at unsignalised crossings
2. PEDESTRIAN SAFETY
Increased awareness of environmental problems and the need for physical fitness encourage the demand for provision of more and better pedestrian facilities. To provide better pedestrian facilities, the appropriate standard and control of the facilities need to be determined. To decide the appropriate standard and control of pedestrian facilities, pedestrian studies, which consist of pedestrian data collection and pedestrian analysis, need to be done. One of the objectives of the pedestrian studies is to evaluate the effects of a proposed policy on the pedestrian facilities before its implementation. The implementation of a policy without pedestrian studies might lead to a very costly trial and error due to the implementation cost (i.e. user cost, construction, marking etc.).
The safest and most effective pedestrian crossings often use several traffic control devices or design elements to meet the information and control needs of both motorists and pedestrians. The desirable characteristics for a pedestrian crossing adopted are given below:
• The street crossing task is made simple and convenient for pedestrians
• The crossing location and any waiting or crossing pedestrians have excellent visibility
• Vehicle speeds are slowed or controlled in the area of the pedestrian crossing
• Vehicle drivers are more aware of the presence of the crosswalk
• Vehicle drivers yield the right-of-way to legally crossing pedestrians
• Pedestrians use designated crossing locations and obey applicable state and local traffic laws
In a complex (e.g., multi-lane, high-speed, high-volume) street environment, it probably will be difficult to provide these characteristics with a single simple treatment, i.e., complex street environments may require several different treatments intended to serve different purposes. For a multi-lane high-volume arterial street, the following treatments could be adopted:
• A median refuge island to make the street crossing easier and more convenient
• Advanced yield lines to improve the visibility of crossing pedestrians
• Removal of parking and installation of curb extensions to improve visibility
• Pedestrian-activated flashing beacons to warn motorists of crossing pedestrians
• Motorist signs to indicate that pedestrians have the legal right-of-way
• Pedestrian signs to encourage looking behaviour, crosswalk compliance, and push button activation
In the unsignalised pedestrian crossing, the following alternatives can be adopted to reduce the pedestrian vehicular conflicts.
• Marked crosswalk
• Enhanced, high-visibility, or “active when present” traffic control device
• Red signal or beacon device; or
• Conventional traffic control signal
2.2 PEDESTRIAN CHARACTERISTICS
The traffic flow characteristics could be divided into two categories, microscopic level and macroscopic level. Microscopic level involves individual units with traffic characteristics such as individual speed and individual interaction. Most of the pedestrian studies that have been carried out are on a macroscopic level. In macroscopic pedestrian data-collection, all pedestrian movements in pedestrian facilities are aggregated into flow, average speed and area module. The main concern of macroscopic pedestrian studies is space allocation for pedestrians in the pedestrian facilities. It does not consider the direct interaction between pedestrians and it is not well suited for prediction of pedestrian flow performance in pedestrian areas or buildings with some street furniture (kiosk, benches, telephone booths, fountain etc). Microscopic pedestrian studies on the other hand, treat every pedestrian as an individual and the behaviour of pedestrian interaction is measured. Though the microscopic pedestrian studies do not replace the macroscopic pedestrian studies, it considers a more detailed analysis for pedestrian facilities and interaction.
In contrast to a pedestrian who walks alone, the increase in the number of pedestrians in the facilities creates problems of interaction. The pedestrians influence each other in their walking behaviour either with mutual or reciprocal action. They need to avoid or overtake each other to be able to maintain their speed, they need to change their individual speed and direction and sometimes they need to stop and wait to give others the chance to move first.
In a very dense situation, they need to maintain their distance/headway toward other pedestrians and surroundings to reduce their physical contact to each other. Thus, a pedestrian tends to minimize the interaction between pedestrians. Because of the interaction, the pedestrians feel uncomfortable, and experience delay (inefficiency). Interaction between pedestrians, as the important point in the microscopic level, can be modelled as a repulsive and attractive effect between pedestrians and between pedestrian with the environment.
The importance of detailed design and pedestrian interaction is best exemplified using the case studies that are described below. Teknomo, K. (2002) conducted a study on microscopic pedestrian flow characteristics which used microscopic pedestrian simulation to determine the flow performance of pedestrians in the intersection of pedestrian malls and doors as illustrated in Fig.1 & Fig.2 below. The intersection of pedestrian malls with roundabout is shown in Fig.1 a. Each pedestrian is denoted by an arrow.
The study compared the flow performance (comfort ability and delay) of the pedestrians in the intersection with and without the round about. It was revealed that pedestrian flow performance of the intersection with round about is better than without the round about. Pedestrian flow that was more efficient could even be reached with less space. Those simulations have rejected the linearity assumption of space and flow in the macroscopic level. Fig.1 b represents two rooms connected with two doors and the pedestrians are coming from both sides of the rooms. Two simple scenarios were experimented. The first scenario was letting the pedestrians pass through any door (two way door), while in the second scenario each door can be passed by only one direction. The result of the experiments showed that one-way door is better than a two-way door. The movement of pedestrian needed to be controlled so that the interaction problem is reduced.
Fig.1 a Pedestrian Movement without Control (Source: Ref.5)
Fig.1 b Pedestrian Movement without Control (2 door movement) (Source: Ref.5)
Fig.2 a Controlled Pedestrian Movement (Source: Ref.5)
Fig.2 b Controlled pedestrian movement (2 door movement) (Source: Ref.5)
2.2.1 Walking Speed
Pedestrians have a wide range of needs and abilities. The Manual on Uniform Traffic Control Devices (MUTCD) includes a walking speed of 4.0 ft/s (1.2 m/s) for calculating pedestrian clearance intervals for traffic signals. MUTCD also saying that, where pedestrians walk more slowly than normal or pedestrians in wheel chairs routinely use the crosswalk, a walking speed of less than 4.0 ft/s (1.2 m/s) should be considered in determining the pedestrian clearance times. Other research studies have identified pedestrian walking speeds ranging from 2.2 to 4.3 ft/s (0.6 to 1.3 m/s).
In 2002, the U.S. Access Board used the guidelines prepared by the Public Rights-of-way Access Advisory Committee and recommended a universal maximum pedestrian walking speed of 3.0 ft/s (0.9 m/s). TCRP/NCHRP study in 2003 showed that the 15th percentile walking speed for younger pedestrians is 3.77 ft/s (1.15 m/s) (sample size of 2,335), and the 15th percentile walking speed for older pedestrians is 3.03 ft/s (0.92 m/s) (sample size of 106). The older pedestrian groups (male, female, and both) had 15th percentile walking speeds that differed statistically from the 15th percentile walking speeds of the younger pedestrians. The data collected with databases contain more than 2,000 pedestrian crossings in 2003 for the TCRP/NCHRP study identified a slower walking speed for the younger group (3.77 ft/s [1.15 m/s]).
For calculating the effects of pedestrian the following walking speeds were considered
• 1.3.5 ft/s (1.1 m/s) walking speeds for general population
• 3.0 ft/s (0.9 m/s) walking speeds for older or less able population
3. PEDESTRIAN VEHICULAR INTERACTION AT UNSIGNALISED CROSSINGS
Pedestrian is often the most vulnerable of all transportation system users, and frequently the most overlooked. Accidents between pedestrians and vehicles are examined in terms of minimizing conflict between the two modes. The pedestrian vehicular interaction is an area of traffic engineering where thorough study is required. The interaction not only leads to the delay or reduction in capacity but also causes accidents. Pedestrian vehicular interaction decides the vehicular flow characteristics especially where there is a large pedestrian crossing exhibit. Therefore the pedestrian behaviour and driver’s attitude to pedestrians are important.
Most of the pedestrian safety depends to a large extent on vehicular speeds. The driver and pedestrian behaviour differs to great extent from country to country and can be explained as good in developed countries when compared to developing countries.
There are several crossing tactics. Fundamentally, the crossing tactics are related to the waiting time at a curb and walking speed adaptation on the crosswalk. Hunt and Williams  categorized crossing tactics into one-stage, two-stage, walk’n-look, and stationary-crossing. In the one-stage tactic, pedestrians cross a roadway all at once from origin curb to destination curb. The two-stage crossing consists of the first stage crossing from origin curb to median or refuge island, and the second stage crossing from median to destination curb. The walk’n-look tactic involves pedestrians walking along the pavement while scanning vehicle flow and cross directly when a suitable gap appears. The stationary-crossing allows pedestrians to cross between stationary vehicles.
3.1 ZEBRA CROSSINGS
A zebra crossing is simply an unsignalised portion of the carriageway where the pedestrian has legal priority over the motor vehicle. The cross strip is outlined by parallel lines of studs and marked with alternate black and white thermoplastic strips parallel to the centerline of the road, the beginning and the end of each crossing are marked by flashing yellow beacons. The first ever experiments using the uncontrolled pedestrian crossing or unsignalised pedestrian crossing was initiated in London in 1927.
There are three situations: first the pedestrian is waiting at zebra crossing and none of the approaching vehicle stops, second the vehicle stops and gives way to pedestrian and third the pedestrian crosses the zebra crossing in spite of approaching vehicle and forces the vehicle to stop. Pedestrian behaviour is also different. Pedestrian can use zebra line or can cross nearer to zebra line.
Ibrahim et al. (2005) conducted a study inside the university of Malaya campus which is used to describe the driver-vehicle interaction. The vehicles were arriving at an average speed of 30 – 40 kmph. The width of the road was 7m and the road was one-way carriageway. Height of the curb/side walk was 20 cm above carriageway.
3.1.1 Driver Behavior at Zebra Crossing
For the study of driver’s attitude towards pedestrian, total vehicles which do not give way to pedestrian while she/he is waiting at zebra crossing to cross were counted. Then the vehicle which stops (if any) to give way to pedestrian was also recorded. In the 3 hours data there were 96 instances when the pedestrian was present at zebra crossing (on side curb) and the vehicle was approaching. In 13 instances pedestrian was waiting at zebra crossing (on the carriageway) and the vehicle was approaching. If a vehicle was at a distance of 30m or less from the zebra crossing while the pedestrian reached the crossing point, this vehicle was not expected to stop for safety reasons. Therefore when the pedestrian reached the crossing point; the vehicles beyond 30m from the zebra crossing were counted.
They found that the probability of vehicles stopping for pedestrian on the carriageway was more than that on the curbside. It was found that 20% would stop, 32 % won't stop while 52% would slow down so that the pedestrian could cross without them needing to stop at the crossing. This showed a wide gap between what the people think and what they do while they drive. The results are alarming and need urgent attention of relevant authorities/agencies to improve the situation, before the zebra crossing becomes a safety hazard.
3.1.2. Pedestrian Behavior at Zebra Crossing
For the understanding of pedestrian attitude towards zebra crossing, first the study is conducted to check the usage of zebra crossing. They found out that in the one hour data, out of 337 pedestrians crossing the road only 56 used the zebra crossing. The other 281 crossed the road from a distance less than 10m from zebra crossing. Two reasons may be attributed to this behaviour; either pedestrian did not realize the importance of crossing the road at zebra or the wrong placement of zebra crossing. It was found that 45% pedestrians felt safe at zebra crossing, 17% felt that any convenient place along the road will do while 21% felt safe at both places and remainder did not feel safe crossing at either place. This again showed a wide gap between what the people think and what they do while crossing the road.
The crossing speeds between genders were compared while crossing the road. It was found that males move at a higher speed than females. The average crossing speed of male was 1.3m/sec, while the average moving speed of female was 1.15m/sec. The male speed was closer to the US Highway Capacity Manual 2000, values for Level of Service ‘A’ while the female speed was in the range of Level of Service ‘D’. The waiting time is the time elapsed between the pedestrian reaches the zebra crossing and the point when she/he starts crossing. More than 80% of the pedestrians had to wait for less than 7 seconds. Because vehicles moved in platoons and there were less willingness to give way to pedestrians, some pedestrians had to wait as long as 23 seconds. Fig.3 given below shows the waiting time distribution for the pedestrians at zebra crossing.
Fig.3 Waiting Time Distribution for the Pedestrians at Zebra Crossing (Source: Ref.3)
3.1.3 Study Results on Zebra Crossings
Traditionally, the essence of zebra crossing on the road is primarily to maintain a peaceful and safe interaction between pedestrian and vehicular traffic, since it has not been possible to maintain a perfect and complete segregation between these two important road users. Ibrahim et al. (2005) conducted a study on pedestrian and driver behavior at zebra crossings and the following conclusions can be drawn from that study:
• The willingness of drivers to give way to pedestrians at zebra crossing was very low
• The motorbike riders were not willing at all to give way to pedestrians
• What the drivers and pedestrians claimed to do was quite different from what they actually did
• There was a significant difference between the speeds of male and female
• The waiting time for most of the crossing pedestrians was quite low (less than 5 seconds) but for some it was as high as 23seconds
• The results of this study were alarming and needed urgent attention of authorities and the government agencies to look into the vulnerable pedestrian safety problem
3.2 PEDESTRIAN VEHICULAR INTERACTION AT ROUND ABOUTS
As the round about is a space sharing system, it produces an interaction during pedestrian crossings between pedestrians and vehicles at a crosswalk area. Interaction affects the behaviour of all system users in their decision process. For instance, when a pedestrian-vehicle collision on a round about crosswalk occurs, injury to the pedestrian is obviously more critical. Another example of the influencing behaviour due to interaction is the right-of-way allocation between vehicles and pedestrians on a crosswalk. The example is based on the assumption that there is no clear mention of right-of-way laws or the system users are either unaware of or misunderstand the law.
Round about reduces vehicle speeds, minimize vehicle weaving, automatically establish right-of-way, and reduce conflict points from 32 to 8 according to the FHWA Round about Guide (2000). The circulatory vehicle movements at round abouts eliminated or drastically reduced the critical conflicts resulting from red light running, left-turns against opposing traffic, right-angle conflicts at corners, and rear-end collisions. As a result, round abouts significantly reduced vehicular crashes. The studies conducted in US revealed that round about reduced all vehicular crashes by 39 percent and injury crashes by 76 percent on 24 intersections that are converted into round abouts later.
3.2.1 Study Results on Round Abouts
Chae et al. (2002) conducted a study on round abouts and found out that when pedestrian demand reached its capacity, people may tried to cross without sufficient vehicle gaps. However, at or below intersection pedestrian capacity, round abouts were likely safer for pedestrians than traditional intersections for three reasons:
• Round abouts can handle the same or higher pedestrian capacity as a traditional intersection
• Round abouts have fewer pedestrian-vehicle conflict points, and
• Any pedestrian crashes would involve lower impact speeds
If pedestrian volumes exceed the intersection pedestrian capacity, special treatments will be needed for pedestrians such as crosswalks, all motorists would be required to yield to pedestrians, and/or pedestrian signals including enunciators. The study included pedestrian gap detection ability and priority give-way behavior at single-lane round abouts. They found that blind pedestrians missed more crossing opportunities and waiting time was at a rate 3 times higher than sighted pedestrians. It was found that drivers frequently yielded to pedestrians on the entry lane but not on the exit lane. At multi-lane round abouts, a study on drivers’ yielding behaviour and pedestrians’ gap detecting ability showed the same tendency as the results at single-lane roundabouts.
3.2.2 Accident study
Conflict analysis is a tool for accident study evaluations. The study conducted by Chae et al. (2005) measured the comparative operation of round abouts and conventional intersections. If there are fewer vehicle-pedestrian conflicts (Fig.4), the intersection should be safer, all other factors being equal.
Fig.4 Conflict Points between Pedestrian and Vehicle in a Round about (Source: Ref.2)
Fig.5 Conflict Points in Conventional Intersection and Round about (Source: Ref.2)
By considering the conflict locations of Fig.5, potential improvements in pedestrian safety estimated by counting the pedestrian vehicle accidents that occurred at or near conventional intersection conflict points. It was eliminated by the roundabout design. The circulatory vehicle movements at round abouts eliminated or drastically reduced the critical conflicts resulting from red light running, left-turns against opposing traffic, right-angle conflicts at corners, and rear-end collisions. As a result round abouts significantly reduced vehicular crashes. It was found that modern roundabouts are safer than other methods of intersection traffic control.
Fig.6 Speed and Accident’s Severity Relation (Source: Ref.3)
When the speed of the traffic stream decreased the severity of pedestrian accidents also decreased. The severity with respect to speed is shown by Fig.6. The speed of the vehicle approaching the intersection in the round about was less than the speed of the vehicle approaching in the case of conventional round abouts. Therefore the severity of pedestrian accidents at the round about intersection would be less than the severity at conventional intersections.
For reducing the pedestrian vehicular interaction and to improve safety measures for pedestrians, the researchers have developed detection systems .With continued research, it is anticipated that the safety of unsignalised pedestrian crossings can be facilitated by using passive pedestrian detection systems. The analysis of this detection systems have shown that they will help in various crossing applications. The study is described below.
4. PASSIVE PEDESTRIAN DETECTION AT UNSIGNALISED CROSSINGS
Installing standard signage and markings provide limited safety improvements for pedestrians. Pedestrians may also tend to develop a false sense of security from these warning devices. Overtime, motorists who travel the areas with these crossings tend to become conditioned to the signs and markings, thus providing no increase in safety for pedestrians. One possible solution to this is a warning device that operates only when a pedestrian is present. Such a device may be a yellow beacon activated by the pedestrian using a push button and a reflective pedestrian crossing sign. This type of device would catch the attention of the motorist who has become conditioned to the crossing by presenting something that is not present at all times, such as the yellow beacon. However, if one observes a crossing with push buttons, one can observe that not all pedestrians use the buttons. Whether this is due to the button being hard to find, poorly located, or not expected or the pedestrian has a vision impairment, it would be beneficial to have the warning devices activated without having to push a button. A better approach for triggering the device may be passive detection of the pedestrian.
4.1 PASSIVE PEDESTRIAN DETECTION
Passive pedestrian detection is to detect the presence of pedestrians in a stationary or moving state at the curbside of and/or in a pedestrian crossing by means other than those requiring physical actuation by the pedestrian. These means may be by using infrared, ultrasonic, microwave radar, video imaging, or piezometric sensors.
Passive pedestrian detection may be one solution in making unsignalised pedestrian crossings safer. Currently, many sensors are being developed and are in use by different industries that may also be used to detect pedestrians passively. Some of these sensor technologies include the following: video imaging, infrared, ultrasonic, piezometric, and microwave radar. The intended use of these sensors is to detect a pedestrian waiting in the landing areas of a crossing (passive detection). Once a pedestrian is detected, a warning device, such as a yellow beacon, would be actuated to warn oncoming motorists of the pedestrian's presence. To ensure that pedestrians can clear the crossing once they have entered it, another sensor can be positioned to detect pedestrians within the crossing itself and prolong the yellow beacon or flashing pedestrian crossing sign until the pedestrian clears the crossing.
4.2 PORTLAND PASSIVE PEDESTRIAN PROJECT
The study was conducted by Beckwith et al. (2009) at the city of Portland, Oregon in US. A passive infrared, microwave radar, and two ultrasonic sensors were chosen to be tested in the Portland passive pedestrian project. Sensors were chosen by using a decision matrix in the form of the Quality Function Deployment (QFD) method to evaluate how well each sensor met the needs of the project. Performance, maintenance, and cost requirements were all analyzed as part of the QFD method. Once sensors that met the needs of the project were found, preliminary and secondary testing was conducted on each of them.
4.2.1 Quality Function Deployment Method
The quality function deployment method was employed in this project to help evaluate sensors and better understand the scope of the project before development and testing began. It also provided a structure of the information for future development of similar systems. The following six steps comprise the QFD method:
• Identify the customer using the product
• Determine customer requirements
• Determine the importance of each requirement to the customer
• Set benchmarks for the existing products
• Translate customer requirements into measurable engineering requirements
• Set engineering targets for the product
A large portion of the information used in the evaluations of the various sensor technologies came from the manufacturers that typically had never tested or used the sensors in applications involving detection of pedestrians in external environments. This required that preliminary (short-term) tests on the sensors be performed to see how well pedestrians could be detected in an external environment. Secondary (long-term) tests were then conducted on sensors that performed well in the initial testing to observe how well they functioned at an actual crossing.
4.2.2 Preliminary (Short-Term) Testing
The main goals here were to determine whether the detectors could detect pedestrians, the types of detection zones that could be expected, the location requirements and whether there were an excessive number of false calls. A few false calls from the sensors are acceptable and can be compensated for through various methods, whereas not detecting pedestrians can be fatal. Therefore, if a sensor could not consistently detect a pedestrian, it was automatically excluded from consideration in the passive detection project. If sensors were found that could consistently detect pedestrians, they were further tested to determine whether false calls were being made and could be reduced.
For this project, a location that showed a high level of pedestrian traffic adjacent to a bus stop was chosen to conduct the preliminary testing of each sensor. Each sensor was mounted on a pedestrian signal and positioned to detect pedestrian traffic passing on the sidewalk. The sensors were then connected to a Type 170 controller at the location. The controller cabinet was retrofitted with two lights mounted on top that illuminated each time a pedestrian entered the detection zone of the sensor. This permitted two sensors to be tested at once. At this point, adjustments were made for aiming, delay settings, and size of detection zones.
The intersection chosen for preliminary testing was equipped with video cameras and video cassette recorder that allowed for monitoring the sensors over extended periods without having an observer present at all times. The cameras were set to monitor the lights mounted on the controller cabinet and the detection zones of each sensor. This made it possible to ensure that the detection zones were not fluctuating due to environmental changes and that pedestrian outside the zones, animals, or wind blowing through trees was not causing false detections.
Each sensor's zone of detection was marked on the ground. This allowed the analyst to determine whether a pedestrian was detected or not detected or a false call was received. Determination of detection was accomplished by observing whether a pedestrian was within a detection zone and observing whether the lights on the controller cabinet were illuminated. The intersection was videotaped at various times during the day. This gave various conditions at different temperatures throughout the day that could affect sensor operations. The sensors also needed to be tested under various weather conditions. Testing of sensors during dark and light hours did not need to be considered here since lighting has no effect on the operation of the chosen sensors.
With the preliminary testing complete, ways to reduce the number of false calls were also researched. Most often, the way false calls were reduced is dependent on the type of sensor. Each type of sensor has its own operating characteristics and adjustments. Some were equipped with delay or sensitivity adjustments. Many other sensors had no adjustments method other than physically moving them from one position to another. Still, others were adjusted through a software package sold with the sensor. The companies that develop and manufacture each sensor can provide insight into the best methods for sensor adjustments to reduce false calls.
4.2.3 Preliminary Test Results
Of the detectors chosen using the QFD method, three were tested: the passive infrared, Doppler radar, and one ultrasonic sensor. Selection of these three sensors was based on their effectively meeting performance, maintenance and cost requirements. The ultrasonic and passive infrared sensors had the greatest limitations on detection distances and had been used little in applications involving pedestrian detection in an external environment. They were, therefore tested at close and extended ranges, with the latter showing how well they operated at their maximum ranges. The Doppler radar was designed and tested only at medium and extended ranges in anticipation that it would be used as an extended range sensor. Table 1 shows preliminary test results.
Table 1.Preliminary Test Results (Source: Ref.1)
Sensor False calls Detection No Detection Total Pedestrians
Ultrasonic at long range 8%
Ultrasonic at close range 8%
Doppler radar 1%
Passive infrared at close range 4%
Passive infrared at long range 4.5%
(The first number is the percentage of pedestrians out of the total number of pedestrians observed at the crossing that falls within each category. The second number is the actual number of pedestrians who were observed that fell into each category. An empty cell means no pedestrians fell into that category.)
4.2.4 Secondary (Long-Term) Testing
From the preliminary testing, two of the three types of sensors were chosen for further testing at an existing pedestrian crossing. The infrared sensor was chosen for monitoring the landing areas of the crossing, and Doppler radar was chosen for monitoring the area within the crossing itself. The infrared sensors had a very good detection rate and were versatile regarding sensor positioning. This allows the detector to be installed in many different types of applications with minimum upgrading required to existing facilities and also low installation time and cost. The Doppler radar sensor was the only sensor that effectively detected pedestrians at a distance of 9.1m (30ft) or greater and had no maximum operating angles. It also had a detection zone that was wide enough to cover the width of a standard crossing. Therefore, only one to two sensors are needed to effectively monitor a crossing, keeping installation time and cost at a minimum.
4.2.5 Installation Site
The intersection of S. W. Naito Parkway and Couch Street in U.S was chosen as the installation site to perform secondary testing on the passive infrared and Doppler radar sensors. At this intersection, the passive infrared sensors were used to detect pedestrians at the landings, and the Doppler radar detected pedestrians within the crossing. The crossing is approximately 22.6 m (74 ft) in length with a 4.0-m-wide (13-ft-wide) pedestrian island in the middle. The sensors actuate yellow beacons placed above reflective yellow pedestrian crossing signs suspended above the crossing. There were advanced warning signs on each approach along Naito Parkway, and the intersection is controlled by a stop sign on the Couch Street approach.
This crossing poses three problems for applying the given sensors:
•The long crossing length means sensors monitoring the crossing, not the landings, must detect for greater distances. This means less overlapping of detection zones for these sensors
•The pedestrian median allows the pedestrian an opportunity to stop at a halfway point in the crossing. This can allow pedestrian detection to be lost
•Traffic is not required to stop. Since the sensors do not sense a difference between pedestrians and automobiles, both will be detected, which potentially means the beacons may be actuated even when pedestrians are not present
A four-sensor configuration was designed to help address these problems and provide the safest possible crossing for the pedestrian using existing facilities. The plan view of pedestrian crossing at S.W Naito Parkway and Couch Street is shown by Fig.7.
Fig.7 Plan view of Pedestrian Crossing at S.W Naito Parkway and Couch Street (Source: Ref.1)
4.2.6 Four-Sensor Configuration
The four-sensor crossing consists of two passive infrared (1 and 4) and two Doppler radar (2 and 3) sensors. As discussed previously, the infrared sensors were positioned above each landing of the crossing. The Doppler radar sensors were positioned to detect pedestrians within the crossings with overlapping detection zones at the halfway point. The logic for this configuration is simple. Sensor 1 or 4 waits in a detect mode to be activated by a pedestrian entering one of their detection zones. Once activated, the yellow beacons turn on, sensors 2 and 3 will wait in a detect mode, and a minimum beacon activation timer (T1) plus a gap timer (T2) will be invoked. During the period the timer (T1) is active, the system checks to see whether sensors 2 and 3 are detecting pedestrians. If pedestrians are detected, then the system continues to check whether the timer (T1) has expired. If it has and pedestrians are still detected within the crossing, then the call to the yellow beacons will remain active if the gap timer (T2) time interval has not expired. If no pedestrians are detected, then the system checks to see whether the gap time has been exceeded with no detection. If it has not, then the system continues to monitor for pedestrians. If there is still no detection of pedestrians, and the gap time expires, then sensors 2 and 3 are turned off, the beacons are turned off, and sensors 1 and 4 are reactivated (see the flow chart in Fig.8). The gap time (6 sec) has been determined as the time it takes a person walking 0.9 m/sec (3 ft/sec) (AASHTO design value for an elderly pedestrian) to cross the widest lane of vehicular travel that is, in this case, 6.4 m (21 ft). The minimum timer (T1) value was derived using the same method used for the gap timer value. It can be varied to an interval greater than or equal to the gap timer. This allows the minimum timer to be adjusted for specific needs of various applications. The previously mentioned gap time is used to provide insurance against intermittent detection by sensors 2 and 3, or a pedestrian stopping on the pedestrian island for a small amount of time. If the sensors lose detection of the pedestrians while they are in the crossing or on the pedestrian island, then the gap time keeps sensors 2 and 3 active even if no pedestrians are detected in the crossing. This crossing had sight distances well above the safe stopping sight distances on all approaches. The predetermined gap time exceeded the time it would take a vehicle traveling 56.3 km/h (35 mph) to stop on wet pavements safely, which is 68.6 to 76.2 m (225 to 250 ft) (10).
Fig.8 Sensor Logic Flow chart (Source: Ref.1)
When sensors 2 and 3 are active, pedestrians and motor vehicles can be detected. This means that during times of heavy traffic volumes, the beacons may stay illuminated for extended periods. This occurs because the gap time is not allowed to elapse due to the heavy traffic being detected by the sensors. Vehicles being detected by sensors 2 and 3 have two advantages. Since there is a pedestrian island at the halfway point, pedestrians will tend to cross halfway and then wait for a gap in traffic to finish crossing. This wait can be extensive during heavy traffic, which means sensors 2 and 3 may lose detection of the pedestrians because they are not moving. However, since the motor vehicles are detected, the gap time is not allowed to elapse, and once the pedestrians move again, they are again picked up by the sensors.
The four-sensor configuration for this crossing becomes a problem if a pedestrian stop on the island for a period greater than the gap time and there is no vehicular traffic present for that same amount of time. In this case, the sensors will deactivate and the beacons will turn off. This means the pedestrian must finish the crossing with no supplementary warning.
4.2.7 Secondary Test Results
Initial secondary test results and observations were obtained from the S.W. Naito Parkway and Couch Street location in U.S. The test site was observed to learn whether detectors were reliably detecting pedestrians and the length of time the beacons remained active after the pedestrian left the crossing. Five items were recorded:
• Weather conditions: a record of the weather conditions under which the sensors have been tested. This is one of the limiting factors for sensor operation
• Date: different times of the year may show variations in pedestrian and motorist traffic patterns
• Time of day: recorded in 15-minute intervals. Different times of day will have different volumes of pedestrian traffic
• Detection reliability: each sensor was observed during a pedestrian’s crossing to check whether it false-detected with no pedestrian present (F), detected a pedestrian with no problems (D), intermittently detected a pedestrian (I), or lost detection of a pedestrian in the crossing (L)
• System shutdown time: a record of how long after the pedestrian leaves the crossing the beacons remain activated. With cars and pedestrians keeping the Doppler radar sensors activated, beacons can be extended long after the pedestrian has left the crossing
Of the 60 crossings observed, there were eight intermittent (I) detections with pedestrians present in the Doppler radar zones and one in the passive infrared zones. At no time during any of the observed crossings were pedestrians not detected or caught within the crossing when the system shut down after the gap time (T2) elapsed.
On an average, beacons would remain activated after the pedestrian left the crossing for 32 sec. This is twice the time needed for a pedestrian walking at 1.22 m/sec (4 ft/sec) (AASHTO design speed for an average pedestrian) to cross the 22.6-m (74-ft) crossing. The maximum time recorded for beacons remaining on was 125 sec with a minimum time of 6 sec. Longer-than-average activation periods can be expected at this crossing during periods of heavy traffic volumes due to passing vehicles keeping sensors activated. The minimum time will not fall below 6 sec since this is the time period set for the gap timer, explained in the previous section.
During heavy rainfall, if the passive pedestrian detection system had been activated by a pedestrian, the Doppler radar sensors would have remained active, keeping beacons illuminated. Once the rain subsided to a light rainfall, then the Doppler radar sensors would have stopped detecting. This can be viewed as either a benefit of the system or a failure, a failure because the system operates even when no pedestrian is present at the crossing; a benefit because beacons will be activated during the adverse weather condition providing supplementary warning of a crossing with possible pedestrians present
5. SUMMARY AND CONCLUSIONS
The reviewed works demonstrated that pedestrian safety at unsignalised crossings can be carried out safely by using passive pedestrian detection systems. The technology developed in the form of a quality function deployment matrix is helpful in selecting the type of the sensors which can be used in passive pedestrian detection systems. At the same time, it provided a structure or record of the information about the equipment that can be used in the current or similar future applications. Many new devices are being developed and old ones improved for use in the detection of pedestrians. The infrared and Doppler radar sensors that passed the preliminary testing discussed in this report have shown encouraging initial secondary test results. The development of these passive pedestrian detection system attempts to reduce the pedestrian vehicular conflicts at unsignalised crossings. Specifically, significant improvements to pedestrian crossing can be accommodated by the studies which include how the pedestrians are behaving in different situations of traffic, the driver’s behaviour, the interactions between them, the methods to reduce the interaction to avoid accidents, capacity studies etc. The thorough studies will not only reduces the accidents but also reduces congestion, delay etc.
1. Beckwith, D., Zaworsky, K., (2009), “Passive Pedestrian Detection at Unsignalised Crossings”, Transportation Research Record 1636, Paper: 98-0725, pp. 96-103.
2. Chae, K., Stone, J., Pillalamarri, S., (2002), “The Effects of Roundabouts on Pedestrian Safety”, The South Eastern Transportation Center.
3. Ibrahim, N., Karim, M., Kidwai, F., (2005), “Motorists and Pedestrian Interaction at Unsignallised Pedestrian Crossing”, Proceedings of the Eastern Asia Society for Transportation studies, Vol.5, pp. 120 – 125.
4. Lalani, N., Lord, D., (2006), “Improving Pedestrian Safety at Unsignallised Crossing”, TCRP Report 112/NCHRP Report 562.
5. Teknomo, K. (2002), “Microscopic Pedestrian Flow Characteristics – Development of an Image Processing Data Collection and Simulation Model”, Ph.D. Thesis, Tohoku university, Japan