The route choices of pedestrians under crowded and non-emergency conditions: Two-route experiments and modeling

To study the mechanism of pedestrians' route choice behaviors under non-emergency conditions, we conducted a series of route choice experiments. Participants were required to choose between two routes. Possible controls, including bottleneck, social distancing, extra reward, were tested in the experiments. Results shows that the bottleneck effect can dramatically influence the route-choice behaviors, whereas the impact of social distancing and reward were much weaker. Five typical logit models, including Binary Logit (BL) model, Mixed Logit (ML) model, Panel Logit (PL) model, Latent Class Logit (LCL) model and Latent Class Logit including Panel effect (LCL-P) model were employed. PL and LCL models performed better in this study, while the results of LCL-P model were the best. This suggests the existence and importance of heterogeneity in route choice behavior. Two classes of pedestrians were identified, with one being comfort-seeking and the other being speed-seeking. ML model did not perform well in this study, which is contrary to some previous studies. All these results could be helpful for understanding the essence of pe-destrians' route choice behaviors.