The analysis on the desired speed in social force model using a data driven approach

Abstract

In this paper, the desired speed, one of the parameters of the social force model, is recognized as having large influences on the performance of the model. However, there is a gap in giving strategies of assigning the desired speed to make the social force model produce reliable results. To bridge the gap, firstly, not only six strategies for assigning the desired speed of the social force model are reviewed, but also three more strategies are proposed for the first time by analyzing the experimental pedestrian flow trajectories. Secondly, a framework, which is used to fairly compare simulation models given the reference data, is extended to compare the overall nine strategies of assigning desired speed of the social force model. Thirdly, the comparisons are conducted in a corridor scenario with three different kinds of typical density of pedestrian flow. Based on the simulation results, it is found that the density of the pedestrian flow acts as an effective indicator for the complexity of pedestrian dynamics. Furthermore, it is shown that the degree of heterogeneity of the desired speed among pedestrians determines the performance of the social force models. As a result, the recommendations are given on choosing the strategies of assigning the desired speed according to the different background of applications.