This study proposes a path-tracking algorithm based on feed-forward (preview distance control) and feedback (LQR, linear quadratic regulator) controllers to reduce heading angle errors and lateral distance errors between a predefined path and an autonomous vehicle. The main objective of path-tracking is to generate control commands to follow a predefined path. The feed-forward control is applied to solve heading angle errors and lateral distance errors in the trajectory caused by curvatures of the road by controlling the steering angle of the vehicle. An LQR was applied to decrease the errors caused by environmental and external disturbances. The proposed algorithm was verified by simulating the driving environment of an autonomous vehicle using a CARLA simulator. Safety and comfort were demonstrated using the test vehicle. The study also demonstrated that the tracking performance of the proposed algorithm exceeded that of other path-tracking algorithms, such as Pure Pursuit and the Stanley Method.

In this study, the behavior of the driver was derived by conducting a crash simulation considering automated vehicle accident conditions using autonomous emergency braking (AEB) and a human body model (HBM). Based on car-to-car intersection accident conditions in the OSCCAR project and the actual accident report, a crash accident case was selected. The base crash scenario was reconstructed by conducting a driving simulation with reference to the selected accident cases. Additional simulations applying AEB are performed. Based on the boundary conditions, a car-to-car crash simulation was performed to derive a crash pulse. This crash pulse and HBM were applied to a simple cabin model for conducting driver behavior analysis. The results confirmed that the head behavior of the driver of the opposing vehicle increased in the lateral direction.