A Natural Lane Change Decision Model Using Extreme Value Theory for Mixed Traffic Flow

Given the prevalence of frequent highway accidents, investigating the potential of connected automated vehicles (CAVs) to enhance both traffic efficiency and safety emerges as a pivotal concern. This study addresses how CAVs can harness connected information for informed decision-making, proposing a novel natural lane-changing model rooted in extreme value theory to accommodate mixed traffic flow. Initially, this research develops acceleration, deceleration, and randomization rules for a cellular automata model of two-lane mixed traffic flow, grounded in an analysis of diverse vehicle behaviors. Subsequently, utilizing extreme value distribution, the model determines the maximum probability of CAVs changing lanes at various distances. The study concludes with a numerical simulation analyzing trajectory-velocity diagrams, average travel time, and average speed across different CAV penetration rates. Results demonstrate the model's efficacy in mitigating traffic risks while significantly improving traffic efficiency and safety.