Spontaneous gait synchronisation in the wild: exploring the effect of distance and level of interaction

Abstract

Gait synchronization in pedestrians is influenced by biomechanical, environmental, and cognitive factors. Studying gait in ecological settings provides insights often missed in controlled experiments. This study tackles the challenges of assessing gait coordination in real-world interactions using a dataset of uninstructed pedestrian trajectories recorded in an underground pedestrian street network. The data are annotated for group relations, interaction levels, and physical contact. The main goals of our study is to devise a method to identify gait synchronisation from trajectory data and to provide an in-depth analysis of social factors affecting gait synchronisation in pedestrian groups. To that end, we first propose a method to extract gait residuals from pedestrian trajectories, which capture motion of the body caused by gait-induced oscillations. We thereafter apply a suite of analytical techniques spanning both frequency and nonlinear domains. Frequency-based methods, including the Gait Synchronisation Index and Cross Wavelet Coherence, quantify the alignment of oscillatory patterns in gait. Complementary nonlinear measures, such as Lyapunov exponents, determinism, and recurrence quantification metrics, offer deeper insights into the dynamical stability and predictability of coupled gaits. Results show that higher social interaction and closer distances enhance gait synchronization, reducing stride frequency variation and increasing stability. Additionally, triad formation and relative positioning are shown to influence synchronisation. Overall, our findings suggest that social interactions shape pedestrian gait coordination, with interaction level and distance being key factors.