Modelling and simulation of pedestrian self-evacuation in Pedestrian Mall

Most tourists are unfamiliar with the layout of Pedestrian Mall (PM). To improve evacuation efficiencies of tourists, self-evacuation is modeled and simulated in the paper. Firstly, we model the propagation and acquisition of information for evacuees. Secondly, the egress selection model for individuals, who are familiar with the egress scenarios (FESer), is constructed by improving the Social Force Model (SFM). The reinforcement learning algorithm is introduced to optimize the desired speed direction. Then evacuation of individuals who are unfamiliar with the egress scenarios (UESer) is analyzed. Finally, we simulate the pedestrians' self-evacuation in PM. The results show that the evacuation time is subject to the sight field of the evacuees and the proportion of the FESer. When the proportion of the FESer is 0.4 or the radius of sight is 3 m, the evacuation time will begin to decrease. And the proposed evacuation model will accurately reproduce the choice of egresses.