Inverse Resource Rational Based Stochastic Driver Behavior Model

Human drivers have limited and time-varying cognitive resources when making decisions in real-world traffic scenarios, which often leads to unique stochastic behaviors that can not be explained by perfect rationality assumption, a widely accepted premise modeling driving presume rationally make maximize their own rewards under all circumstances. To explicitly address this disadvantage, study presents novel driver behavior model aims capture the resource stochasticity of human driver's realistic longitudinal scenarios. The principle provide theoretic framework better understand cognition processes human's internal mechanisms as utility maximization subject limitations, represented finite preview horizons context driving. An inverse rational-based reinforcement learning approach (IRR-SIRL) is proposed learn distribution planning horizon cost function with given series demonstrations. A nonlinear predictive control (NMPC) used generate driver-specific trajectories using learned distributions driver. simulation experiments are carried out demonstrations gathered from driver-in-the-loop simulator. results reveal variety