Abstract

In this paper, we propose a decision making algorithm intended for automatedvehicles that negotiate with other possibly non-automated vehicles inintersections. The decision algorithm is separated into two parts: a high-leveldecision module based on reinforcement learning, and a low-level planningmodule based on model predictive control. Traffic is simulated with numerouspredefined driver behaviors and intentions, and the performance of the proposeddecision algorithm was evaluated against another controller. The results showthat the proposed decision algorithm yields shorter training episodes and anincreased performance in success rate compared to the other controller.

 

Quick Read (beta)