Quantifying emergent leadership within pedestrian crowd under laboratory conditions

Even in the absence of social affiliations, asymmetrical influence within pedestrian crowds can emerge, leading to the formation of emergent leadership. However, this phenomenon has rarely been addressed in contrast to the well-documented pre-defined leadership in previous literature. In this study, the transfer entropy (TE) method was employed to quantify emergent leadership using time-lapse motion data of pedestrians under controlled laboratory conditions. The study focused on three typical pedestrian traffic scenarios: unidirectional flow, turning flow, and four-directional flow. In general, TE measures the reduction in uncertainty in predicting a pedestrian's next movement state by knowledge of past movement state of another pedestrian. The asymmetric information transferred within the crowd was inferred. The results show that emergent leadership is particularly pronounced in areas where pedestrians change their direction of movement. Specifically, leadership was predominantly observed at the turn in an L-shaped corridor and around the central obstacle in a cross-shaped corridor. Furthermore, the leader-follower relationship was found to be subtly and dynamically evolving over time. The primary contribution of this study is offering novel insights and an information-theoretic scheme to quantify emergent leadership and understand the nonlinear dynamics of pedestrians' interactions in laboratory settings.