Evaluating pedestrian interaction preferences with a game theoretic autonomous vehicle in virtual reality

Localisation and navigation of autonomous vehicles (AVs) in static environments are now solved problems, but how to control their interactions with other road users in mixed traffic environments, especially with pedestrians, remains an open question. Recent work has begun to apply game theory to model and control AV-pedestrian interactions as they compete for space on the road whilst trying to avoid collisions. But this game theory model has been developed only in unrealistic lab environments. To improve their realism, this study empirically examines pedestrian behaviour during road crossing in the presence of approaching autonomous vehicles in more realistic virtual reality (VR) environments. The autonomous vehicles are controlled using game theory, and this study seeks to find the best parameters for these controls to produce comfortable interactions for the pedestrians. In a first experiment, participants’ trajectories reveal a more cautious crossing behaviour in VR than in previous laboratory experiments. In two further experiments, a gradient descent approach is used to investigate participants’ preference for AV driving style. The results show that the majority of participants were not expecting the AV to stop in some scenarios, and there was no change in their crossing behaviour in two environments and with different car models suggestive of car and last-mile style vehicles. These results provide some initial estimates for game theoretic parameters needed by future AVs in their pedestrian interactions and more generally show how such parameters can be inferred from virtual reality experiments.     

Investigating temporal variations in pedestrian crossing behavior at semi-controlled crosswalks: A Bayesian multilevel modeling approach

“Semi-controlled” crosswalks are unsignalized, but have clear pavement markings and “yield to pedestrian” signs. At these locations, pedestrians and motorists frequently interact to determine who should proceed first. When interacting with drivers, pedestrian crossing decisions are complex events that involve a variety of human responses, as well as vehicle dynamics, traffic characteristics, and environmental conditions. In addition, these complexities can be subject to temporal effects. Without considering temporal variations in pedestrian-motorist interaction, statistical methods could lead to biased coefficient estimates and inaccurate conclusions.The study developed a Bayesian multilevel logistic regression (BMLR) model to capture heterogeneities in pedestrian interaction behavior during four different time periods. The proposed method incorporates time-specific effects that vary randomly between time-periods based on a weakly informative prior. The results indicate significant factors, some of which confirm previous research and some that are new ways to explain pedestrian behavior at the individual level. The identification of variables such as FlowOn and FlowWait sheds light on the interactions between pedestrians – providing more information than the single GroupSize measure.Some consequent safety implications are discussed from the perspectives of vehicle dynamics, vehicle flow rate and pedestrian volume. The more detailed metrics developed in this paper will provide a valuable starting point. for the design of crosswalk controls that will foster a higher degree of compliance and less delay.     

Exploring the relationships between pedestrian behaviours and traffic safety attitudes in six countries

The majority of cross-cultural research on traffic safety has investigated driver behaviour, yet in most low- and middle-income countries, where the weight of the road fatality burden is felt, motorisation rates are significantly lower than in higher-income countries. As such, this approach necessarily excludes large parts of the populations in those settings. In order to investigate the link between traffic safety attitudes and road user behaviours, this study used a self-report measure of pedestrian behaviour, applying it in six countries; Bangladesh, China, Kenya, Thailand, the UK, and Vietnam. Focus was on the relationships between a respondent’s attitude towards risky or rule violating on-road behaviours (of other road users, or more generally, not specific to pedestrians), and the extent to which they reported performing three types of risky pedestrian behaviours (i.e., intentional rule violations, errors in judgement or memory, and aggressive behaviours). Data from a sample of 3,423 individuals was subjected to a series of regression models, revealing significant links between attitudes and pedestrian behaviours in all countries, in all three behavioural factors, after controlling for age, gender, and exposure to the road environment. Differences were found between countries in the strength of these relationships, with weaker connections between attitudes and behaviours in Kenya, and stronger connections in China (with other countries in-between the two). Results are discussed in terms of the need to understand the relationships between social cognitive constructs in the specific country in which a road safety intervention is intended to be implemented.     

Factors related to the use of pedestrian bridges in university students of Honduras

Considering that deaths and injuries due to run overs near pedestrian bridges (PB) are preventable events, this research aims to analyze the factors related to the use of pedestrian bridges in students of a public university of Honduras. This was made through a quantitative methodology; 330 questionnaires were applied to a sample of university students. Results suggest that 17.27% of the respondents don’t use the bridge. Most respondents (62.42%) believe it’s faster to cross the street and that using the bridge is tiresome (79.90%). Pedestrian bridges are also considered to be unsafe places, prone to robberies. Pearson’s r coefficient determined that this belief is significantly associated with the perception of cleanliness, quality of the infrastructure and illumination. There is a correlation between stress and considering the use of the bridge as a tiresome activity. A logistic regression model determined that significant variables explained the use of the PB included: being in a hurry, believing it’s faster to cross the street, thinking that PBs have a bad infrastructure, laziness and thinking that using the PB is tiresome, all of them reducing the chance that someone will use the bridge. On the other hand, non-significant variables in the model included: age, considering that using the PB is stressful, thinking that crossing the street is stressful, thinking PBs have good illumination. Safety related variables were not able to discriminate between users and non-users. These results are discussed in the light of previous research and their implication for public policy and future research.     

Analyzing pedestrian behavior when crossing urban roads by combining RP and SP data

Considering the high crash rates involving pedestrians on urban roads, it is highly relevant to understanding pedestrian crossing behavior. This paper is the first to combine stated preference (SP) and revealed preference (RP) data to evaluate the impact that individual attributes, trip characteristics, built environment, strategies to prevent unauthorized crossing, and traffic flows have on pedestrians crossing decisions in an urban context. SP and RP surveys were designed and collected in Barranquilla (Colombia) near pedestrian bridges or signalized intersections where direct crossings and a high concentration of pedestrian fatalities related to traffic accidents exist. A logit model was estimated using the data enrichment paradigm. Results show that pedestrians weigh risks and costs when choosing how to cross the road. The trajectories observed in the RP component suggest that people prefer direct crossings; nevertheless, pedestrian bridges and signalized intersections can be attractive alternatives if their location matches the origin or destination of the crossing, and no detour is needed to use them. Waiting time; safety; the fine imposed for jaywalking; personal security, and previous decisions are also variables that influence pedestrian behavior when crossing urban roads. These results can be helpful to urban planners and decision-makers interested in proposing appropriate pedestrian infrastructure. The data pooling technique and the inclusion of a cost-related variable (i.e., fine) allowed computing the willingness to pay and marginal substitution rates for attributes of the built environment and other characteristics associated with the crossing decision. Also, the inclusion of several crossing alternatives and situations allowed assessing pedestrian crossing preferences under different scenarios.     

A method in modeling interactive pedestrian crossing and driver yielding decisions during their interactions at intersections

Investigating pedestrian crossing and driver yielding decisions should be an important focus considering the high risks of pedestrians in exposed to motorized traffic. Limitations, however, exist in previous studies – variables considered previously have been limited; how their behavior affect each other (defined as interactive impacts) were not sufficiently considered. This paper aims to provide a methodological approach for pedestrian crossing and driver yielding decisions during their interactions, considering of different variable types including interactive impact variables, traffic condition variables, road design variables, and environment variables. A Distance-Velocity (DV) framework proposed in an earlier study is introduced for definitions and concepts in studying pedestrian-vehicle interactions. Logistic regression, support vector machines, neural networks and random forests, are introduced as candidate models. A case study involving six crosswalk locations is conducted, focusing on interactions between pedestrians and right-turn vehicles. The proposed methodological approach is applied, with the performance of the four machine learning methods compared in terms of model generalization and confusion matrix. The model with the best performance is further compared to the typical gap-based model. Results show that random forest and logistic regression models performed the best in modeling pedestrian crossing and driver yielding decisions respective, in terms of model generalization. Besides, the DV-based modeling method (average accuracy of over 90% for pedestrians and 80% for drivers) outperformed the traditional gap-based method in all test seeds. As a key finding, interactive impacts from each other (the pedestrian and the driver) act as a key contributing variable on their decisions.     

Evaluating the influence of approaching vehicles on pedestrian’s visual patterns and crossing behaviors at an uncontrolled crosswalk

The visual behaviors and movement characteristics of pedestrians are related to their surrounding potential safety hazards, such as approaching vehicles. This study primarily aimed to investigate the visual patterns and walking behaviors of pedestrians interacting with approaching vehicles. Field experiments were conducted at two uncontrolled crosswalks located at the Cuihua and Yanta roads in Xi’an, China. The visual performance of pedestrians was assessed using the eye tracking system from SensoMotoric Instruments (SMI). Moreover, motion trajectories of the pedestrians and approaching vehicles were obtained using an unmanned aerial vehicle. Subsequently, the visual attributes and movement trajectories of pedestrians and motion trajectories of approaching vehicles were statistically analyzed. The results showed that approaching vehicles distracted the fixation of crossing pedestrians significantly, and occupied 29.5% of the total duration of fixation; that is, pedestrians always directed more fixation points to the approaching vehicles compared to other stimuli. As a vehicle approached, pedestrians’ fixation shifted from other areas of interest to the vehicle. Moreover, an increase in the velocity of the vehicle and a closer distance between pedestrian and the vehicle resulted in an increase in the pedestrians’ duration of fixation on the approaching vehicle, and they implemented more saccades. Furthermore, approaching vehicle’s velocity and distance between pedestrian and approaching vehicle are not significantly associated with pedestrian’s movement attributes. These findings provide insights into the crossing behavior of pedestrians during pedestrian-vehicle interactions, which could assist future researchers and policy makers.     

Pedestrians’ road crossing decisions and body parts’ movements

This study aims to examine pedestrians' crossing decision, body parts' movement and full body movement, just before and during road crossing in a simulated setup. To accomplish this, a novel experimental setup for analyzing pedestrians’ crossing behavior and motion was developed where the simulated display was synchronized with a 3D motion capturing system. Twenty participants, divided into control and an experimental time pressure group, observed sixteen short (less than 30 s) and long road (70 s or more) crossing scenarios with varying crossing opportunities. Based on the crossing opportunities they were asked to cross a 3.6 m wide one-lane one way urban road. It was found that the crossing initiation process consists of four incremental movements of body parts: the head and the shoulder first; the hip, wrist and elbow second; the knee as a separate joint, and finally the ankle. Results showed that pedestrians’ decision to cross and body parts movement are influenced by time pressure and wait time for a safe crossing opportunity. Specifically, pedestrians prepare their body parts earlier, initiate their crossing earlier, and adjust their speed to compensate for the risk taken in less safe or non-safe crossing opportunities. Within the control group, women tended to be more risk avoiding than men, however those differences disappeared in the time pressure group. Most importantly, the findings provide initial evidence that this novel simulation configuration can be used to gain precise knowledge of pedestrians’ decision-making and movement processes.     

The estimation of vehicle speed and stopping distance by pedestrians crossing streets in a naturalistic traffic environment

The ability to estimate vehicle speed and stopping distance accurately is important for pedestrians to make safe road crossing decisions. In this study, a field experiment in a naturalistic traffic environment was conducted to measure pedestrians’ estimation of vehicle speed and stopping distance when they are crossing streets. Forty-four participants (18–45 years old) reported their estimation on 1043 vehicles, and the corresponding actual vehicle speed and stopping distance were recorded. In the speed estimation task, pedestrians’ performances change in different actual speed levels and different weather conditions. In sunny conditions, pedestrians tended to underestimate actual vehicle speeds that were higher than 40 km/h but were able to accurately estimate speeds that were lower than 40 km/h. In rainy conditions, pedestrians tended to underestimate actual vehicle speeds that were higher than 45 km/h but were able to accurately estimate speeds ranging from 35 km/h to 45 km/h. In stopping distance estimation task, the accurate estimation interval ranged from 60 km/h to 65 km/h, and pedestrians generally underestimated the stopping distance when vehicles were travelling over 65 km/h. The results show that pedestrians have accurate estimation intervals that vary by weather conditions. When the speed of the oncoming vehicle exceeded the upper bound of the accurate interval, pedestrians were more likely to underestimate the vehicle speed, increasing their risk of incorrectly deciding to cross when it is not safe to do so.     

Distracted pedestrians in crosswalks: An application of the Theory of Planned Behavior

Thousands of adult pedestrians are killed or injured each year. In spite of ever growing knowledge of correlates of risk pedestrian behavior, much remains to be learned. A small body of research has applied the Theory of Planned Behavior (TPB) to the problem of risky crossing decisions. We employed TPB to examination of intentions to cross under varying types of distraction. Eighty adults completed measures of injury exposure, followed by a questionnaire structured around TPB assessing intentions to cross streets under common conditions of self-imposed distraction. Results were similar to other research, suggesting attitude and perceived behavioral control predicted intentions to cross streets while distracted. Results are discussed in the context of similar studies, and with regard to future directions and injury prevention efforts.