Online Robust-Diagnostic System and Method for Decisions of Autonomous Vehicles based on Dynamic Path

Abstract

The safety of self-driving vehicles is always a priority. The right decision is the biggest key to the safety of self-driving cars. Any wrong decision can cause collisions, overturns, and even casualties. In this paper, a diagnostic system and model to evaluate on-road decisions are presented. A future dynamic trajectory of the automatic vehicle, and the parameters of the traffic environment, such as the curvature and width of a traffic lane and the distance to a barrier, are applied to construct the diagnostic model for making the self-driving car has the last line of defense to prevent danger. The improvement of decisions along with our proposed model is investigated through an on-road experiment. The results show that the performance of the proposed framework is reliable enough for robust autonomous driving.