Model-Based Analysis of Real-world Forward Collision Conflicts Using a Closed-loop Sliding-mode Driver Model

摘要

This paper develops a driver control model appropriate to hard-braking events seen in real-world forward collision conflicts.A nonlinear controller is used to represent the actions of a subject vehicle driver,in response to the braking actions of the lead vehicle.The control law provides a target deceleration that includes nonlinear switching based on a sliding mode error variable,plus linear feedback of range-rate and a bias term based on brake light detection.The form of the control law is derived from and closely related to the classical sliding mode controller.From parameter tuning it is possible to achieve a high degree of correlation with measured vehicle kinematics.Through parametric variation,it is shown that collision boundaries exist as nonlinear functions of driver control parameters and reaction timing.This motivates the future use of closed-loop event simulation to provide robust crash surrogate metrics and collision risk estimation.