Lateral Stochasticity in Lane Changing via Logistic Diffusion Process

Lane changing is a complex behavior within traffic flow, as it encompasses the decision-making process of drivers and the interplay among neighboring vehicles. Accurately describing and predicting lane changing behavior is critical for the successful implementation of autonomous vehicles. Although various approaches have been employed to address this task, current models suffer from several drawbacks, such as a lack of consideration for stochasticity and an inability to capture characteristic moments during the lane changing process. This chapter focuses on the real-time estimation of lane changing decisions and the identification of characteristic moments. To achieve this objective, a logistic diffusion model is developed, utilizing a stochastic differential equation. The marginal distribution of the lateral location, which solves the Fokker–Planck equation, is derived and simulated using the Monte Carlo method. Two characteristic moments, namely the Cross-Lane-Mark Moment and lane changing duration, are numerically determined. The results demonstrate that the proposed model efficiently estimates lane changing decisions and accurately derives the two characteristic moments.