Pedestrian movements during large crowded events naturally consist of different modes of movement behaviour. Despite its importance for understanding crowd dynamics, intermittent movement behaviour is an aspect missing in the existing crowd behaviour literature. Here we analyse movement data generated from nearly 600 Wi-Fi sensors during large entertainment events in the Johan Cruijff ArenA football stadium in Amsterdam. We use the state-space modeling framework to investigate intermittent motion patterns. Movement models from the field of movement ecology are used to analyse individual pedestrian movement. Joint estimation of multiple movement tracks allows us to investigate statistical properties of measured movement metrics. We show that behavioural switching is not independent of external events, and the probability of being in one of the behavioural states changes over time. In addition, we show that the distribution of waiting times deviates from the exponential and is best fit by a heavy-tailed distribution. The heavy-tailed waiting times are indicative of bursty movement dynamics, which are here for the first time shown to characterise pedestrian movements in dense crowds. Bursty crowd behaviour has important implications for various diffusion-related processes, such as the spreading of infectious diseases.

大型拥挤事件期间的行人运动自然由不同的运动行为模式组成。尽管它对于理解人群动态很重要，但间歇性运动行为是现有人群行为文献中缺失的一个方面。在这里，我们分析了阿姆斯特丹 Johan Cruijff Arena 足球场大型娱乐活动期间近 600 个 Wi-Fi 传感器生成的运动数据。我们使用状态空间建模框架来研究间歇性运动模式。来自运动生态学领域的运动模型用于分析个体行人运动。多个运动轨迹的联合估计使我们能够研究测量运动指标的统计特性。我们表明行为转换并不独立于外部事件，并且处于其中一种行为状态的概率随时间而变化。此外，我们表明等待时间的分布偏离指数分布，并且最适合重尾分布。重尾等待时间表明突发运动动态，这是这里第一次显示出在密集人群中表征行人运动的特征。突发人群行为对各种与传播相关的过程具有重要意义，例如传染病的传播。这是这里首次展示了密集人群中行人运动的特征。突发人群行为对各种与传播相关的过程具有重要意义，例如传染病的传播。这是这里首次展示了密集人群中行人运动的特征。突发人群行为对各种与传播相关的过程具有重要意义，例如传染病的传播。