Integrating acoustic-based risk quantification for modeling one-to-one dynamic pedestrian-vehicle interactions at non-signalized crosswalks using empirical dataOA

Perceived risk is a key factor influencing pedestrian-vehicle interactions.Existing risk assessments primarily quantify conflict probabilities,overlooking the subjective and ambiguous nature of perceived risk,which differs significantly from precise probabilities.Establishing a mapping between objective probabilistic risk and subjective perceived risk is therefore crucial.Drawing on sound perception theory,this study introduces the concept of decibel(dB)to propose a novel method for quantifying perceived risk while incorporating uncertainty and ambiguity.Combined with evolutionary game theory(EGT),the model analyzes dynamic interactions between pedestrians and vehicles at unsignalized crosswalks,quantifying perceived risks and determining acceptable levels.The acceptable perceived risk range is 1.13-2.48 dB for vehicles and 0-1.85 dB for pedestrians,with pedestrians more likely to yield at lower perceived risks.Perceived risk significantly impacts four evolutionary strategies and their paths.“Pedestrian Wait,Vehicle Not Yield”exhibits steep slopes and rapid convergence,reflecting high-conflict scenarios.“Pedestrian Cross,Vehicle Yield”is stable,with steady convergence,while“Pedestrian Wait,Vehicle Yield”and“Pedestrian Cross,Vehicle Not Yield”are unstable,showing low slopes and weak convergence.Simulations reveal that increasing vehicle distance,reducing speed,and limiting acceleration promote“Pedestrian Cross,Vehicle Yield”,while reducing pedestrian distance and speed shifts behavior toward“Pedestrian Wait,Vehicle Not Yield”.The proposed model and strategy analysis provide valuable insights for designing traffic safety facilities and improving autonomous driving algorithms,enabling real-time monitoring and enhancing pedestrian safety.