The self-slowing behavioral mechanism of pedestrians under normal and emergency conditions

Abstract We study the similarities and differences between the headway-velocity relations under normal and emergency conditions to explore whether they can be described by a unified behavioral equation. We firstly performed a series of pedestrian experiments in three different scenes under normal and emergency conditions respectively to obtain the behavioral parameters of headway-velocity relations based on the visual hindrance field as well as the high precise trajectories of pedestrians. The strong similarities in the headway-velocity relations in both normal and emergency conditions suggest that a unified behavioral mechanism is at play in human-driven pedestrian traffic. This mechanism is essentially a safety-driven self-slowing behavior that pedestrians try to adopt a safe speed for a given spacing between them to avoid collisions and preserve their personal space. We notice that even in emergency escape situations, people still tend to slow down the speed and show defensive behavior when the headway approaches a critical value in order to decrease the collision level and protect the body from contact as much as possible. Moreover, the differences between normal and emergency conditions are also found in the experiments. Compared to normal condition, the free velocity of pedestrians is markedly higher while the minimum critical headway is much shorter in emergency situation. Especially, the proportionality constant, the reciprocal of which is safe response time, is higher under emergency condition. That means pedestrians will slow down the free speed to zero in a shorter safe response time in emergencies. A modified social force model is then proposed to incorporate this self-slowing behavioral mechanism, and different self-slowing behavioral parameters in the model under normal and emergency conditions both refer to the observed experimental data. Simulations with the same setup as the experiments were carried out for both normal and emergency conditions using the unified model with self-slowing, and the simulated spacetime diagrams as well as stop-and-go waves in circular movement under normal condition, the simulated evacuation efficiency in room evacuation under emergency condition and the simulated velocity profiles in corridor scene under normal and emergency conditions all demonstrate remarkable consistency with the experimental results.