Decision Making Process of Autonomous Vehicle with Intention-Aware Prediction at Unsignalized Intersections

In an environment of intelligent transportation systems, interaction with other drivers via identifying their intentions becomes a challenging and unavoidable problem for the driverless vehicles. In this paper, we propose a human-like decision model for unsignalized intersections. An intention-aware prediction for other drivers via a CNN method with multiple-object tracking and Kalman-Filter operations is considered in the developed model to construct the interaction of other drivers. A human-like decision model according to the moving intention of the obstacle was proposed to generate the strategy of an autonomous vehicle. Our approach method is validated on a real autonomous vehicle in the presence of human-driven vehicles through an on-road test.