Decision-Making Model of Autonomous Driving at Intersection Based on Unified Driving Operational Risk Field

Abstract

Safety and comfort are the two major requirements for the successful implementation of self-driving cars, which are anticipated to constitute the future generation of transportation. To create safe and effective self-driving car trajectories, a novel behavioral decision model is developed. First, a risk field model for driving activities based on vehicle kinematics and Eulerian solenoids is constructed. From there, the principle of least action is applied to produce the best trajectory points. Finally, nine typical unit scenarios are simulated by matlab’s driving scenario designer to verify the feasibility of the decision-making algorithm. This study illustrates how an unified operational risk field can efficiently increase intersection passing efficiency while ensuring safety, utilizing the principle of least action. The experimental results show that in the scenario of unprotected left turn and more than 5 vehicles in the intersection, the decision-making model improves the pass rate by 23% compared with the TTI (Time To Intersection) threshold method.