Pedestrian Interaction Research Articles

Simulation of pedestrian flows through queues: Effect of interaction and intersecting angle

In this paper, we propose a microscopic pedestrian model with discrete space representation to simulate pedestrian flows through multiple queues. In the model, two types of pedestrians, i.e., passing pedestrians and queuing pedestrians, are involved and they can intersect in different angles. Both the interaction and movement of pedestrians are jointly formulated by using a potential field. By simulation analyses, we investigate the effects of both the intersecting angle and the interaction between pedestrians on the movement efficiency of pedestrians. The results indicate that (1) the moderate avoidance of queuing pedestrians can result in a high movement efficiency of pedestrians; (2) the phenomenon of passing pedestrians moving near one side of the walkway can be observed in the scenario; and (3) some measures can be taken to improve the movement efficiency of pedestrians in the scenario, such as separating the movement areas of the two types of pedestrians or setting up a set of obstacles near service windows. This paper is also an attempt to study the effect of moving obstacles on the movement of pedestrians.

Pedestrian interaction with automated vehicles at uncontrolled intersections

Automated Vehicles (AVs) are being developed rapidly and tested on public roads, but pedestrians’ interaction with AV is not comprehensively understood or thoroughly investigated to ensure safe operations and the public’s trust of AVs. In this study, we aimed to provide another research evidence to enhance such understanding with the use of external interfaces for facilitating the interaction between pedestrians and AVs. We developed five external interfaces, including text, symbol, animated-eye, a combination of text and symbol, and speed. These interfaces communicated five types of information, including (1) intent of AV; 2) advice to pedestrians of what to do, (3) AV’s awareness of pedestrians, (4) combination of intent and advice, and (5) vehicle movement (i.e., speed). We tested the interfaces through two field studies at uncontrolled intersections with crosswalks. The Wizard of Oz method was used, in which an experimenter worked as a driver in an instrumented vehicle and wore an outfit to be invisible to the pedestrians, thus rendering the set-up to simulate an AV interacting with a pedestrian. The interfaces were displayed on an LED panel mounted on the AV. Results showed that the AV’s external interface did not change pedestrians’ response time in comparison with the baseline without any interface. There was no statistically significant difference in response time among the external interfaces either. According to the post-experimental interview, vehicle movement pattern (e.g., vehicle speed) continued to be a significant cue for pedestrians to decide when to cross the intersections. Participants perceived the communication of the AV’s intent and vehicle speed as more beneficial than the communication of AV’s awareness. The subjective ratings showed positive effects of those interfaces that were easy to understand (e.g., text interface and speed interface), which also helped pedestrians feel safer when interacting with the AV.

EGAT: Extended Graph Attention Network for Pedestrian Trajectory Prediction.

To improve foresight and make correct judgment in advance, pedestrian trajectory prediction has a wide range of application values in autonomous driving, robot interaction, and safety monitoring. However, most of the existing methods only focus on the interaction of local pedestrians according to distance, ignoring the influence of far pedestrians; the range of network input (receptive field) is small. In this paper, an extended graph attention network (EGAT) is proposed to increase receptive field, which focuses not only on local pedestrians, but also on those who are far away, to further strengthen pedestrian interaction. In the temporal domain, TSG-LSTM (TS-LSTM and TG-LSTM) and P-LSTM are proposed based on LSTM to enhance information transmission by residual connection. Compared with state-of-the-art methods, the model EGAT achieves excellent performance on both ETH and UCY public datasets and generates more reliable trajectories.

Automated Vehicle Behavior Design for Pedestrian Interactions at Unsignalized Crosswalks

A significant challenge for real-world deployment of automated vehicles (AVs) is their safe interaction with other road users, particularly pedestrians, whose behavior can change almost instantaneously. To ensure safety, AVs are expected to drive cautiously around pedestrians. However, conservative AV behavior can encourage pedestrians to be careless. Additionally, riders inside the AV expect a comfortable ride and to reach their destinations without excessive delay (AV performance). These objectives of safety, performance, and comfort are opposing in nature and thus have to be balanced in a way that is accepted by both the riders and the pedestrians. For example, in the case of pedestrian crossing at an unsignalized crosswalk, the AV might want to prioritize comfort normally and gradually slow down to stop, but at a school zone might want to prioritize safety and have a stronger deceleration. Existing approach of fixing the weights of the objectives can be constrictive as in reality, the objectives can vary depending on the situation. To overcome this limitation, we incorporated a hybrid systems pedestrian model that is interactive and depends on the vehicle behavior (distance and speed). Our findings suggest the possibility of characterizing driving behaviors by varying the weights of the driving objectives. We believe that by understanding the objective performance of the different driving behaviors with an interactive pedestrian model, the AV-decision making can be expressed by varying the weights of the objective function. The primary contribution of this research is the objective evaluation of a spectrum of driving behaviors represented by the different weight combinations of the three AV driving objectives – safety, performance, and comfort.

Modelling Pedestrians & Cyclists Interaction with Traffic Flow in Urban Areas.(Dept.C)

Little work has been done on pedestrians and cyclists as a viable form of transport in urban areaes to assess their characteristics and their interaction with traffic. Software designed for motor vehicles and are not appropriate for modeling pedestrians and cyclists sharing the traffic this road space in busy shopping areas within the (C.B.D) of towns. There are distinct incompatibilities between traffic characteristics and pedestrians & cyclists behaviour. In this paper an attempt has been made to assess the interaction of both pedestrians and cyclists with traffic flow through a study in urban area (Assiut City C.B.D). Recommended values for pedestrians and bicycles equivalency factors are given for planning applications to enable the best utilization of transport facilities.

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.

Modeling Vehicle–Pedestrian Interactions using a Nonprobabilistic Regression Approach

Understanding how vehicle drivers and pedestrians interact is key to identifying countermeasures that improve the safety of the interactions. As a result, techniques that can be used to evaluate the effectiveness of safety countermeasures and traffic control devices without the need to wait for the availability of crash data are needed. Using video, the interactions between right-turning vehicles and conflicting pedestrians were documented and quantified using vehicle and pedestrian position timestamps. Interactions documented were purposely narrow in scope to obtain a controlled dataset. Logged timestamps enabled the calculation of values such as time to complete a right turn and time for a pedestrian to reach a critical conflict point when a vehicle initiated a right turn. A nonprobabilistic regression model explaining the relationship between the calculated values was created. The model described the expected behavior of right-turning drivers: when drivers perceive the possibility of a pedestrian reaching a critical conflict point at the same time as them, they will modify their behavior, even if not coming to a complete stop. This behavior is not a surprise and has been previously documented in the literature. The primary contribution of this research is demonstrating that by analyzing a narrow set of interactions, clear and simple models that mostly explain the interactions between right-turning vehicles and pedestrians can be developed using nonprobabilistic linear regression techniques. An argument is made that the model parameters can be used to evaluate the effectiveness of traffic control devices.

Impact of stochastic representations of pedestrian actions on serviceability response

Over the past 15 years, there have been some research outcomes in other disciplines that could be used to produce new, more accurate and realistic numerical models to characterise pedestrian loads and to significantly improve predictions of response for multiple-pedestrian scenarios. However, the disconnection between fields has not facilitated this further research. Using this, the paper presents (a) a new sophisticated load model that includes a description of vertical and lateral loads, including pedestrian–structure interaction, (b) the numerical description of relationships to describe the key parameters of the proposed model and (c) the evaluation of the effects of pedestrian characteristics that are relevant for serviceability response of footbridges. The proposed new load model allows for the inherent variability of individual pedestrian actions (intra-subject variability), a probabilistic description of how pedestrian characteristics vary among subjects (inter-subject variability) and collective human behaviour (pedestrian–pedestrian interaction). Some of these characteristics are not currently considered in design approaches and can have a substantial impact on structural response assessments. Finally, recommendations are made for many of these characteristics to be introduced in analyses to evaluate the vibration serviceability limit state of footbridges in a more accurate and realistic manner.

Thermodynamics of a gas of pedestrians: theory and experiment

In this paper, we perform an experiment on the interaction of pedestrians in a chaotic environment and investigate the possibility to study its results using a thermodynamic model. In contrast to simple single-file unidirectional scenarios, where only distance and time are relevant to adjust walking speed, bidirectional cases are much more complex since pedestrians can perform evading manoeuvres to avoid collisions. To better understand collision avoidance in a bidimensional environment we designed a set of experiments where people need to move chaotically for the whole time. Trajectories of moving pedestrians were obtained by tracking their head position, but a method to obtain body orientation failed, thus limiting reliable information on average quantities, i.e. average density and speed. By analysing those data, we showed that equations for thermodynamic processes can be used to describe pedestrian dynamics from medium densities or a state where interaction distances are very small. To allow combining low density cognitive aspects of collision avoidance with semi-random motion at medium densities we also developed a microscopic simulation model inspired by physics. Results show that, after calibrations, the simulation model allows to reproduce the fundamental diagram of different studies despite the very simple rules implemented. This shows that describing the statistical nature of specific crowds requires a relatively small set of rules and research should focus on cognitive/psychological aspects which are essential for understanding crowds of people.

Follower-Leader Concept in Microscopic Analysis of Pedestrian Movement in a Crowd

This paper presents a microscopic analysis of factors influencing pedestrian movement and interactions with their surroundings for two considered modes: independent movement influenced only by the surrounding conditions and synchronized movement based on following another pedestrian. This study analyses which of these effects prevail in different phases of the movement. The results show that the significant value of correlation between pedestrian velocity and corresponding individual density is observed mainly during approaching the crowd. Contrarily, in the segment of pedestrian trajectory which corresponds to movement inside the crowd, correlation between the velocity of a follower and a leader is more important. This confirms that the pedestrian behaviour in a crowd is a complex field.

Game Theory-Based Framework for Modeling Human–Vehicle Interactions on the Road

New application domains have faded the barriers between humans and robots, introducing a new set of complexities to robotic systems. The major impediment is the uncertainties associated with human decision making, which makes it challenging to predict human behavior. A realistic model of human behavior is thus vital to capture humans’ interactive behavior with their surroundings and provide robots with reliable estimates on what is most likely to happen. Focusing on operations of connected and automated vehicles (CAVs) in areas with a high presence of human actors (i.e., pedestrians), this study creates an interactive decision-making framework to predict pedestrians’ trajectories when walking in a shared environment with vehicles and other pedestrians. It develops a game theoretical structure to approximate the movement and directional components of pedestrian motion using the theory of Nash equilibria in non-cooperative games. It also introduces a novel payoff structure to address the inherent uncertainties in human behavior. Ground truth pedestrian trajectories are then used to calibrate the game parameters and evaluate the model’s performance in approximating the motion decisions of human agents in interaction with interfering vehicles and pedestrians. The main contribution of the study is to develop an interactive human–vehicle decision-making framework toward realizing human–vehicle coexistence by capturing the effect of pedestrian–vehicle and pedestrian–pedestrian interactions on choice of walking strategies. The derived knowledge could be used in CAV navigation algorithms to provide the vehicle with more accurate predictions of pedestrian behavior, and in turn, improve CAV motion planning in human-populated areas.

Modeling the interactions of pedestrians and cyclists in mixed flow conditions in uni- and bidirectional flows on a shared pedestrian-cycle road

The mixed flow of pedestrians and cyclists is frequently observed on the roads they share, but investigations of the dynamics of this kind of mixed flow have been very limited. This study proposes a heuristic-based model to reproduce the mixed-flow dynamics of pedestrians and cyclists, and the model is calibrated with an experiment on the mixed traffic flow of pedestrians and cyclists. Pedestrians/cyclists were asked to walk/ride on a ring-shaped track. In the uni/bidirectional flow scenario, pedestrians and cyclists moved in the same/opposite direction. A genetic algorithm was used for parameter calibration. The model could reproduce the experimental results well. Under both scenarios, pedestrians and cyclists formed their own lanes. The pedestrians walked in the inner lane, and cyclists rode in the outer lane in a self-organized process. The widths of the pedestrian lane and the cyclist lane were found to be more uniform during bidirectional flow. The pedestrian flow rate was higher in the unidirectional flow scenario than in the bidirectional flow scenario. In contrast, at low cyclist densities, the cyclist flow rate was essentially the same in both scenarios. When the density was high, the cyclist flow rate is higher in the unidirectional flow scenario. Sensitivity analyses showed that cyclist speed had little effect on the pedestrian flow rate. A higher cyclist speed led to a higher cyclist flow rate at low densities, but the cyclist flow rates approached the same value at high cyclist densities. As the proportion of pedestrians/cyclists increased, the flow rate of cyclists/pedestrians decreased. The simulation results on a straight track were largely consistent with those on a ring-shaped track.

Comparison of Methods to Evaluate the Influence of an Automated Vehicle’s Driving Behavior on Pedestrians: Wizard of Oz, Virtual Reality, and Video

Integrating automated vehicles into mixed traffic entails several challenges. Their driving behavior must be designed such that is understandable for all human road users, and that it ensures an efficient and safe traffic system. Previous studies investigated these issues, especially regarding the communication between automated vehicles and pedestrians. These studies used different methods, e.g., videos, virtual reality, or Wizard of Oz vehicles. However, the extent of transferability between these studies is still unknown. Therefore, we replicated the same study design in four different settings: two video, one virtual reality, and one Wizard of Oz setup. In the first video setup, videos from the virtual reality setup were used, while in the second setup, we filmed the Wizard of Oz vehicle. In all studies, participants stood at the roadside in a shared space. An automated vehicle approached from the left, using different driving profiles characterized by changing speed to communicate its intention to let the pedestrians cross the road. Participants were asked to recognize the intention of the automated vehicle and to press a button as soon as they realized this intention. Results revealed differences in the intention recognition time between the four study setups, as well as in the correct intention rate. The results from vehicle–pedestrian interaction studies published in recent years that used different study settings can therefore only be compared to each other to a limited extent.

Pedestrian and cyclist flows interaction in the urban street and road network

Pedestrian and cyclist flows interaction in the urban street and road network

A Model to Predict Children’s Reaction Time at Signalized Intersections

Traffic accident statistics in urban areas, both locally in Croatia and at the European level, identify children as a group of vulnerable road users. The analysis of the parameters that influence the interaction of child pedestrians and other road users requires special attention. This paper presents the results of research about the reaction time of children, measured both in laboratory conditions, via a computer reaction time test, and in actual traffic conditions. The results of the reaction time test in a situation with expected stimuli (a computer test) of children aged 6 to 10 years were compared with the results of the reaction time of adult traffic participants, drivers, who also took part in the computer test. Standard deviations of the reaction times between the control group (drivers, adults) and each subgroup of children were significantly different (p < 0.05). The results suggest that the largest developmental jump occurs between preschool children and first-grade children. In actual traffic conditions, the reaction time of children aged 4 to 16 years at the signalized intersection was measured. The model for predicting the reaction time of children in real traffic conditions was created using a neural network. The model prediction results matched well with the values measured in actual traffic conditions, for the observed intersection (correlation coefficient is 94.56%) and for the validation intersection (correlation coefficient is 92.29%). Parameters influencing children’s reaction times in real traffic conditions were identified by applying both statistical analysis and the neural network model developed. Using both methods, the same key distractors were identified—the movement of children in the group and the use of mobile phones. The case study was conducted at selected signalized intersections in the city of Osijek, Croatia.

Craniofacial Trauma Associated with the Introduction of Electric Scooters in an Urban Setting

With the increase in popularity of standing electric scooter (e‐scooter) alternatives to public and automobile transit comes the potential for an added burden of traumatic injuries by unsafe, unregulated, or unconventional use of these modes of point‐to‐point travel. The ergonomics and riding position of e‐scooters make riders particularly susceptible to fall trauma, and the lessened ability to break falls can lead to an increased incidence of craniofacial trauma. When complicated by the variable laws and regulations of different urban areas utilizing rental e‐scooters, the mass usage of this expedient form of transport exposes users to a potential public health risk. The purpose of the present study was to evaluate emergency department (ED) admissions directly linked to e‐scooter use to assess the incidence and types of craniofacial trauma directly over the first 7 months of e‐scooter rentals in a major urban center.A total of 90 patients (56 males, 34 females) were admitted to the ED because of e‐scooter–related injuries over the course of the study, ranging in age from 13 to 60 years (mean age, 31.8 years). Most (64.4%) admissions involved extremity injuries, with 18 fractures, 4 dislocations, and 10 lacerations. Eight patients (8.8%) required transfer to the intensive care unit, male patients experienced trauma at a greater rate (62.2%) of e‐scooter–related injuries, and 7 of the patients admitted (7.7%) were younger than 18 years. Alcohol consumption was recorded in 16 e‐scooter–related ED visits (17.8%), and none of the admitted patients reported the use of protective headgear. 52 (57.7%) individuals presented with craniofacial trauma related to e‐scooter use. Of the 52 patients, 28 had trauma to the upper face, 28 to the midface, and 9 to the lower face. Soft tissue facial injuries were noted in 44 patients (80.8%) who sustained craniofacial trauma, 28 (53.8%) of whom required laceration closure. Extensive craniofacial injuries included subarachnoid/ subdural hemorrhage in 6 patients (11.5% of craniofacial patients), 1 of whom required emergency craniotomy. Le Fort fractures occurred in 3 patients, including one bilateral Le Fort II/III and left‐sided Le Fort I fracture and two Le Fort I fractures.The present study found that more than one half of all trauma patients presenting to a Level I trauma ED in Dallas, Texas, for e‐scooter injuries had injuries to the head and face. Craniofacial injuries require specialist intervention ranging from dentoalveolar splinting to extensive facial reconstruction, highlighting the implications of craniofacial trauma risk associated with e‐scooter use that could be significantly reduced by the use of proper protective equipment and access to well‐regulated bicycle lanes and cyclist/pedestrian safety policies. Special focus on the factors contributing to traumatic events involving e‐scooters, such as helmet use, pedestrian interaction, alcohol use, and policies regarding scooter use will help establish the best practices for the mitigation and treatment of such injuries.

Surrogate safety indicator for unsignalised pedestrian crossings

ProblemAlthough the road safety situation in Poland is generally improving, the number of accidents at pedestrian crossings has not decreased in the last four years. This paper presents the results of the MOBIS research project, the aim of which was to develop surrogate safety indicators, based on detection of pedestrian-vehicle conflicts using video analysis. MethodPedestrian and vehicle traffic was filmed at two unsignalised pedestrian zebra crossings in Warsaw and Wrocław for over 40 days. Motion trajectories of vehicles and pedestrians were determined based on video processing. Several variables describing pedestrian-vehicle interactions were calculated, such as speed, post-encroachment time, distance between the participants, decelerations, etc. Classification of encounters was based on interactions of pedestrians and vehicles i.e. drivers yielding to pedestrians, vehicles passing just in front of, or behind pedestrians. Results and discussionCriteria for identification of dangerous encounters were selected with the assumption that it should be possible to automate the assessment process. The selected variables were: pedestrian-vehicle passing distance and the vehicle speed at that moment. Other criteria were used in cases of abrupt braking – deceleration exceeding 4 m/s2 and vehicle speed. A Dangerous Encounter Index is proposed as a surrogate safety indicator for pedestrian crossings. It relates the occurrence of dangerous events to exposure, defined as the number of pedestrian-vehicle encounters. Practical applicationsThe proposed index shows that crossing two lanes involves more risk than crossing one lane, given similar traffic flow. Some improvement of safety at both types of crossing was observed after active signage involving blinking lights had been introduced. The proposed method is a step towards automation of safety assessment.

Simulation of simultaneous departure from Hiroo Gakuen using a pedestrian interaction model

Simulation of simultaneous departure from Hiroo Gakuen using a pedestrian interaction model

Light-Based External Human Machine Interface: Color Evaluation for Self-Driving Vehicle and Pedestrian Interaction

In today’s road traffic pedestrians seek eye contact with drivers to move along safely. Such communication is no longer possible with self-driving vehicles. Previous research shows that an external Human-Machine-Interface (eHMI) provides an interface between self-driving vehicles and pedestrians. However, recommendations for standardization are still being developed. The study compares the colors white and turquoise for eHMI lamps indicating that the automated driving system is engaged. The colors are evaluated in a street-crossing scenario and a parking lot scenario with a Wizard-of-Oz vehicle equipped with eHMI lamps mounted on top of the vehicle. We conducted questionnaires and structured interviews with N=59 participants to identify eHMI design guidelines. Our research provides evidence that turquoise facilitates pedestrians’ factors like visibility, discriminability, sense of safety and trust higher than white. The results are consistent among traffic scenarios. This paper contributes in formulating research-based design guidelines to improve pedestrian safety.

Context Model for Pedestrian Intention Prediction Using Factored Latent-Dynamic Conditional Random Fields

Smooth handling of pedestrian interactions is a key requirement for Autonomous Vehicles (AV) and Advanced Driver Assistance Systems (ADAS). Such systems call for early and accurate prediction of a pedestrian's crossing/not-crossing behaviour in front of the vehicle. Existing approaches to pedestrian behaviour prediction make use of pedestrian motion, his/her location in a scene and static context variables such as traffic lights, zebra crossings etc. We stress on the necessity of early prediction for smooth operation of such systems. We introduce the influence of vehicle interactions on pedestrian intention for this purpose. In this paper, we show a discernible advance in prediction time aided by the inclusion of such vehicle interaction context. We apply our methods to two different datasets, one in-house collected - NTU dataset and another public real-life benchmark - JAAD dataset. We also propose a generic graphical model Factored Latent-Dynamic Conditional Random Fields (FLDCRF) for single and multi-label sequence prediction as well as joint interaction modeling tasks. FLDCRF outperforms Long Short-Term Memory (LSTM) networks across the datasets ($\sim$100 sequences per dataset) over identical time-series features. While the existing best system predicts pedestrian stopping behaviour with 70\% accuracy 0.38 seconds before the actual events, our system achieves such accuracy at least 0.9 seconds on an average before the actual events across datasets.