Lessons learned from pedestrian-driver communication and yielding patterns

Understanding the hidden patterns of tacit communication between drivers and pedestrians is crucial for improving pedestrian safety. However, this type of communication is a result of the psychological processes of both pedestrians and drivers, which are very difficult to understand thoroughly. This study utilizes a naturalistic field study dataset and explores the hidden patterns from successful and failed communication events using a pattern recognition method known as Taxicab Correspondence Analysis (TCA). The successful communication scenarios indicate the combinations of variable attributes such as eye contact, facial expression, the assertion of crossing, and effective traffic control devices are strongly associated with successful scenarios. The patterns for failed scenarios are most likely to be on the roadway with a relatively higher speed limit (e.g., 35 mph) and a relatively lower speed limit (e.g., 15 mph) under different conditions. On roadways with a higher speed limit, the failed scenarios are highly associated with passive and undecisive pedestrians, pedestrians far away from the crosswalk, regardless of pedestrian-driver eye contact and facial expression of the pedestrians. Instead of waiting for pedestrians to making a crossing decision, overspeeding drivers are more likely to speed up and pass the crosswalk. On roadways with a lower speed limit, the failed scenarios are often associated with distracted pedestrians, vehicles having the right of way, and the absence of effective traffic control devices. These findings could help transportation agencies identify appropriate countermeasures to reduce pedestrian crashes. The findings on driver-pedestrian communication patterns could provide scopes for improvement in computer vision-based algorithms designed for autonomous vehicle industries.     

Pedestrians’ road crossing behaviour in front of automated vehicles: Results from a pedestrian simulation experiment using agent-based modelling

The objective of this research is to explore the relation between personal characteristics of pedestrians and their crossing behaviour in front of an automated vehicle (AV). For this purpose, a simulation experiment was developed using Agent-Based Modelling (ABM) techniques. Sixty participants were asked to cross the road in a virtual environment displayed on a computer screen, allowing to record their crossing behaviour when in the presence of AVs and conventional vehicles (CVs). In some experimental configurations, the AVs communicated their intention to continue or not to continue their trajectories through the use of lights. The ABM allowed controlling the behaviour of the vehicles when interacting with the simulated avatar of the respondents. The subjects of the experiment were also asked to fill in a questionnaire about usual behaviour in traffic, as well as attitudes and risk perceptions toward crossing roads. The questionnaire data were used to estimate individual specific behavioural latent variables by means of principal component analysis which resulted in three main factors named: violations, lapses, and trust in AVs. The results of generalized linear mixed models applied to the data showed that besides the distance from the approaching vehicle and existence of a zebra crossing, pedestrians’ crossing decisions are significantly affected by the participants’ age, familiarity with AVs, the communication between the AV and the pedestrian, and whether the approaching vehicle is an AV. Moreover, the introduction of the latent factors as explanatory variables into the regression models indicated that individual specific characteristics like willingness to take risks and violate traffic rules, and trust in AVs can have additional explanatory power in the crossing decisions.     

Pedestrian detection using a moving camera: A novel framework for foreground detection

While background subtraction techniques have been widely applied to detect moving objects in a video stream captured by a static camera, detecting moving objects using a moving camera still represents a challenging task. In this context, pedestrian detection using a camera placed on the top of a vehicle’s windshield has been rarely investigated. This is mainly due to the background ego-motion. Since the scene captured by the camera seems in motion, it is very difficult to distinguish the moving pedestrians from the others that belong to the static part of the scene. For this reason, a compensation step is needed to suppress the ego-motion. This paper presents a study on the main challenges facing pedestrian detection systems as well as methods proposed to handle these challenges. A novel trajectory classification framework for detecting pedestrians even in challenging real-world environments is proposed. The proposed method models the background motion between two consecutive frames in order to compensate the camera motion. Then, it defines a classification process that differentiates between the background and the foreground in the frame. Using the defined foreground, we consequently identify the presence of pedestrians in the scene. The proposed method was validated on a public benchmark dataset: CVC-14 containing both visible and far infrared video sequences in day and night times. Experimental results confirm the effectiveness of the proposed approach in capturing the dynamic aspect between frames and therefore detecting the presence of pedestrians in the scene.     

Racial bias in driver yielding behavior at crosswalks

Psychological and social identity-related factors have been shown to influence drivers’ behaviors toward pedestrians, but no previous studies have examined the potential for drivers’ racial bias to impact yielding behavior with pedestrians. If drivers’ yielding behavior results in differential behavior toward Black and White pedestrians, this may lead to disparate pedestrian crossing experiences based on race and potentially contribute to disproportionate safety outcomes for minorities. We tested the hypothesis that drivers’ yielding behavior is influenced by pedestrians’ race in a controlled field experiment at an unsignalized midblock marked crosswalk in downtown Portland, Oregon. Six trained male research team participants (3 White, 3 Black) simulated an individual pedestrian crossing, while trained observers cataloged the number of cars that passed and the time until a driver yielded. Results (88 pedestrian trials, 173 driver-subjects) revealed that Black pedestrians were passed by twice as many cars and experienced wait times that were 32% longer than White pedestrians. Results support the hypothesis that minority pedestrians experience discriminatory treatment by drivers at crosswalks.     

An examination of the influence of crosswalk marking removal on pedestrian safety as reflected in road user behaviours

In some local authorities in Israel, road markings were removed from non-signalised crosswalks situated on multilane divided urban roads. This study examined whether this treatment contributes to pedestrian safety, based on the analysis of differences in road user behaviours and pedestrian crossing conditions at sites without crosswalk markings (treatment sites) versus those with the markings (comparison sites). Field observations were conducted at two treatment and four comparison sites, including video-recordings and free-flow speed measurements. The speed data analysis indicated that treatment sites were associated with higher speeds as compared to other sites, at least in one travel direction, and that speed values were sufficiently high to create a threat of severe injury for the crossing pedestrians. Examinations of other road user behaviours showed that treatment sites are associated with higher obedience of pedestrians to safe crossing rules and low rates of pedestrian-vehicle conflicts during the crossing, but also with long waiting times by pedestrians prior to crossing, low rates of giving-way to pedestrians and more vehicles ignoring pedestrian needs to cross. In general, pedestrian crosswalk removal brought about a worsening in the possibility of crossing the road for pedestrians that in combination with high vehicle speeds cannot be recognised as a safety-improving measure. The study did not support the claim that removing a marked crosswalk leads to safety benefits for pedestrians.     

Revisiting the level-of-service framework for pedestrian comfortability: Velocity depicts more accurate perceived congestion than local density

The evaluation of pedestrian comfortability is important for the construction and management of walkable spaces. Pedestrian level-of-service (LOS), which is mostly categorized by density, has been widely applied and is believed to be capable of indicating the comfortability of the crowd. However, there is a lack of evidence that physically measured LOS reflects psychological comfortability, let alone the comparison between the performances of different candidate indicators, including the velocity and different local densities. Here, we show that walking velocity depicts pedestrian perceived congestion more accurately than density. In our experiments on pedestrians in a room-egress scenario with an inner obstacle, we obtained objective physical trajectory data from video analysis as well as subjective perception data from a questionnaire survey. The performance of the velocity and four types of local densities in reproducing pedestrian perceptions were numerically evaluated. We found that the velocity outperformed the local density. The lower performance of local densities was mainly caused by the pedestrians located at the back of the crowd, who walked at lower velocities and perceived higher congestion, despite their lower local densities that would correspond to less crowdedness from a physical viewpoint. Besides, the perceived congestion of pedestrians was shown to be affected by the initial pedestrian positions at the crowd, the obstacle layout, and the pedestrians’ age and gender. Furthermore, we suggest that the larger the gap between the desired and actual velocities, the larger the extent of the perceived congestion. We expect that our findings will contribute to a more accurate evaluation of pedestrian comfortability, which could help improve the walkable spaces of various infrastructures.     

Predicting pedestrians’ intention to cross the road in front of automated vehicles in risky situations

Due to the absence of a human driver, the introduction of fully automated vehicles (FAVs) may bring new safety challenges to the traffic system, especially when FAVs interact with vulnerable road users such as pedestrians. To ensure safer interactions between pedestrians and FAVs, this questionnaire-based study aims to understand Australian pedestrians’ intention to engage in risky road-crossing behaviors when they interact with FAVs vs. human-driven vehicles (HDVs). A 2 × 2 between-subject design was utilized, in which two risky road-crossing scenarios were designed and took into account the vehicle type (FAV vs. HDV) and vehicle speed (30 km/h vs. 50 km/h). A total of 493 participants (aged 18–77) were randomly assigned to one of the four experimental conditions and completed an online questionnaire based on the extended Theory of Planned Behavior (TPB). This questionnaire measured pedestrians’ intentions to cross the road in the assigned scenarios as well as the motivational factors behind these intentions in terms of attitude, subjective norm, perceived behavioral control, perceived risk and trust in the vehicle. The results show that pedestrians had significantly higher intentions to cross the road in front of approaching FAVs than HDVs. Participants also reported a lower risk perception of crossing in front of FAVs and greater trust in this type of vehicle. Attitude, subjective norm, and perceived behavioral control were significant predictors of intentions to engage in risky road-crossing behavior. Findings of this study provide important implications for the development and implementation of FAVs in the future road transport system.     

Evaluation of motorized two-wheeler rider responses towards jaywalking pedestrians through mockup control studies for urban streets

This study aims to model the motorized two-wheeler (MTW) riders’ evasive-action behavior towards jaywalking pedestrians using a mockup study. The brake reaction times (BRTs), approach speeds, decelerations, headings, and yaw rates were analyzed for two surprise scenarios (scenarios 1 and 3), one stationary scenario (scenario-2), and one expected scenario (scenario-4). In total, 50 riders participated in the mockup study. The results revealed that the 90th percentile BRT for the expected and surprise scenarios were 3.6 and 1.6 s, respectively. Further, repeated-measures ANOVA was performed followed by mixed effect modeling to ascertain the effect of conflict severity (two groups: group-1 with Time to Collision (TTC) < 1.5 s and group-2 with TTC > 1.5 s) and scenario type (three groups: scenarios 1, 3 and 4) on BRT. The results indicated that the main effects were significant while the interaction effect was not significant. The positive and significant coefficient (0.32) of TTC group-2 indicated higher BRTs than TTC group-1. Considering scenario-1 as the base scenario, the coefficient of scenario 3 (-0.02) indicated that scenario-1 and scenario-3 had a similar effect on BRT, while the coefficient of scenario-4 (1.47) indicated higher BRTs compared to scenario-1. The analysis of evasive action behavior revealed that 32% of riders performed hard braking in surprise scenarios. Further, yaw rate values at the crossing point indicated a loss of control of MTW in 90% of surprise events. The observations from this study provide a basis for developing countermeasures to improve pedestrian and MTW rider safety.     

Estimating effectiveness of speed reduction measures for pedestrian crossing treatments using an empirically supported speed choice modeling framework

Pedestrian crossing treatments, such as Pedestrian Hybrid Beacons (PHBs) and Rectangular Rapid-Flashing Beacons (RRFBs), are traffic control devices implemented to help pedestrians safely cross busy or higher-speed roadways at midblock crossings and uncontrolled intersections. This study aims to assess the effectiveness of the PHB and RRFB by analyzing drivers’ speeding behavior under different roadway types with real-life traffic conditions. In order to understand the effect of pedestrian crossing treatments (i.e., PHB and RRFB) have on drivers’ speeding behavior, this paper analyzes four zones (i.e., one upstream zone and three consecutive downstream zones). For the four analysis zones, different indexes were computed which corresponds to the differences in drivers’ speed when encountered with the pedestrian crossing treatments. A grouped random effect hurdle beta regression model is estimated for the indexes, with a fixed effect hurdle beta model used for comparison and validating the importance of considering the grouped random heterogeneity across participants. The proposed analysis framework was validated by the means of an empirical driving simulator study, based on two urban roads in the Central Florida region, North Alafaya Trail (SR-434) and South Orange Blossom Trail (US-441). The results revealed that the proposed modeling framework reflects drivers’ difference in speed for the different pedestrian crossing treatments. The results suggest that with proper understanding of the PHB, the PHB can reduce drivers speed even beyond the location of the PHB. Meanwhile, the RRFB does have some effect in speed reduction beyond the location of the RRFB, however many drivers failed to acknowledge the RRFB. It is suggested that when drivers’ have proper education on the use of the PHB to reduce speed safely and for the installation of RRFB be on roads with two or less lanes and a speed limit less than 40 mph. While the main purpose of the pedestrian crossing treatments is to help pedestrians cross safely, speed reduction can be considered a byproduct as revealed in this study.     

The influence of a pedestrian countdown display on pedestrian behavior at signalized pedestrian crossings

Pedestrian behavior has been a subject of surveys carried out at two signalized pedestrian crossings, in the city of Doboj (BIH, Republic of Srpska). The analysis is made on the basis of video recordings, and includes behavior at pedestrian crossings, in various conditions (two locations, different vehicle flows, with or without a countdown display). The analysis also includes the distribution of illegal crossings (at the pedestrian red light) of different categories of pedestrians (gender, age), before and after the installation of a pedestrian countdown display, at two different pedestrian crossings. The results of the study have shown that a countdown display reduces statistically significantly the total number of violators, regardless of its location and traffic flow. However, this reduction is not the same at various categories of pedestrians (male, female, elderly, young people). The influence of the display has been particularly dominant at the pedestrian crossing located outside the centre of the city, with the smaller intensity of traffic. Children pedestrians do not accept the behavior in accordance with the countdown display in case of a reduced intensity of traffic, at a pedestrian crossing located in the backstreet. A countdown display does not statistically significantly change the distribution form of the number of offences during the red light for pedestrians, regardless of the pedestrian environment. However, there is a statistically significant difference in certain intervals (during the first and last 4 s) in the total distribution, as well as in the categories (female pedestrians, children). A countdown display does not reduce the overall number of “slow” pedestrians, but the impact on certain categories of “slow” pedestrians (gender/children), as well as on pedestrian violators, per age categories, varies depending on the location and traffic flow at an intersection.