Data-Driven Crowd Evacuation Simulation Method Based on Raspberry Pi

Abstract

When public emergencies or disasters occurred, such as fires and earthquakes, safety accidents with group trampling can easily be caused. In the previous work, the study of crowd evacuation mainly adopted artificial parameters for simulations. To overcome these problems, we propose a data-driven crowd evacuation simulation method based on Raspberry Pi. First, we configure Raspberry Pi as sensor node, which is connected the wireless WiFi network card with monitoring function and the infrared sensor module with detection function. And then the sensor node was deployed to collect real data of crowd evacuation in real scenarios. Second, the curve fitting method is used to deal with the functional relationship between crowd density and average evacuation speed. In terms of crowd evacuation simulation calculation, we present Crowd Density-based Reciprocal Velocity Obstacle (CD-RVO) algorithm which drive individual speed changes with crowd density, to optimize the options for safety exit of individuals and reduce evacuation time effectively. Finally, the experimental results show that the proposed method can simulate the process of crowd evacuation realistically and improve crowd evacuation efficiency.