Microscopic interaction behavior between individual pedestrians

Due to growing mobility in densely populated urban areas, public pedestrian walkways are becoming more and more crowded. Since space in these densely occupied areas is highly valued, the use of every square meter must be evaluated carefully. Microscopic pedestrian simulation models can provide such evaluation. However, fundamental knowledge on individual pedestrian behavior underpinning these models is hardly available. This study focuses on the so-called interaction behavior between individual pedestrians. It comprises the identification of aspects that influence individual interaction behavior, their effects on pedestrian movement and how these effects can be described and predicted. Assessment criteria and pedestrian experiments In unhindered walking, personal walking capabilities determine the intended path in lateral and in longitudinal direction. Walking outside of this intended path in unhindered walking is defined as interaction movement. Also, the passing side and the extent of evasion from the mean intended path are investigated. The effects of the direction of approach (i.e. bidirectional, crossing and overtaking situations), individual gender and motion pace and small groups on these three assessment criteria are examined. Laboratory experiments have been performed to obtain data on individual pedestrian interaction behavior. In these experiments twelve participants are initially asked to walk a length of 20 meters unhindered by other pedestrians. Thereafter, two or three participants are asked to walk simultaneously, in such a way that interaction between them is provoked. This has resulted in 1586 trajectories of individual pedestrians. Analyses on unhindered walking and interaction movements The unhindered trajectories are used to derive boundaries for the intended path in unhindered walking, mean walking paths and mean walking speeds. In addition, a method is presented to analyze the swaying effect in unhindered walking. It has been shown in this study that the observed step lengths and step frequencies are inversely proportional to each other. It is found that individual pedestrians perform interaction movement in 88% of the interaction situations observed in this study. These movements can be interpreted as some gallantry towards other pedestrians. This finding justifies the detailed modeling of pedestrian interaction movements in microscopic simulation models. In the bidirectional situation pedestrians strongly prefer passing each other on the right hand side. In crossing situations, walking in a hurry increases the probability of passing in front of other pedestrians and meeting a small group of two pedestrians increases the probability of passing at the back of this group. The passing side and the direction of approach mainly determine the direction and the extent of evasion from the individual mean walking path and from the individual mean walking speed during interaction. Pedestrians seem to prefer larger lateral evasion in bidirectional situations and larger longitudinal evasion in crossing situations. Men laterally evade more than women and hurried pedestrians laterally evade more than normally walking pedestrians. It seems that hurried pedestrian are either granted ‘right of way’ or that they take the initiative themselves in an early stage of the interaction process. The extent of evasion is larger when small groups are encountered. Implications for pedestrian modeling and recommendations for further research The observations on pedestrians in individual interaction situations differ in a number of issues from observations on individual pedestrians in flows. This indicates that behavioral reactions of individual pedestrians to other individual pedestrians differ from behavioral reactions of individual pedestrians to pedestrian flows. Moreover, it follows from this study that the passing side determines the ensuing lateral and longitudinal evasion. A conceptual model for implementing this into pedestrian models is given in the attached figure. This model differs from other models since interaction movements are determined based on a decision on the passing side instead of the other way around. Furthermore, hurried pedestrians behave significantly different than normally walking pedestrians and this aspect should thus be modeled. Several aspects regarding individual pedestrian interaction need further research. Among these are the initiation moment of interaction movements and the effects of gender, body size and age on pedestrian interaction. These aspects can be investigated with the observations obtained in this study. The effects of multiple individual pedestrians on the individual interaction behavior (i.e. individual behavior within pedestrian flows) were not investigated in this study. To gain knowledge on these effects it is of interest to analyze whether behavioral reactions of individual pedestrians to other individual pedestrians significantly differ from behavioral reactions of individual pedestrians to pedestrian flows.Transport & PlanningCivil Engineering and Geoscience