Automated vehicles (AVs) are deemed to be the key element for the intelligent transportation system in the future. Many studies have been made to improve the AVs' ability of environment recognition and vehicle control, while the attention paid to decision making is not enough though the decision algorithms so far are very preliminary. Therefore, a framework of the decision-making training and learning is put forward in this paper. It consists of two parts: the deep reinforcement learning (DRL) training program and the high-fidelity virtual simulation environment. Then the basic microscopic behavior, car-following (CF), is trained within this framework. In addition, theoretical analysis and experiments were conducted on setting reward function for accelerating training using DRL. The results show that on the premise of driving comfort, the efficiency of the trained AV increases 7.9% compared to the classical traffic model, intelligent driver model (IDM). Later on, on a more complex three-lane section, we trained the integrated model combines both CF and lane-changing (LC) behavior, the average speed further grows 2.4%. It indicates that our framework is effective for AV's decision-making learning.

No abstract

Adaptive cruise control (ACC) system, as one of the most fundamental modules of automated vehicles, is widely used in commercially available vehicles. It inevitably influences the traffic flow, from both the individual perspective, that is, its interaction with other traffic participants, and the traffic system perspective, that is, traffic string stability and road capacity. However, subject to limited data availability, no consistent conclusions on these impacts have been reached in the literature. Meanwhile, the similarities and differences between ACC vehicles and human-driven vehicles (HDV) have not been fully discussed and comparisons among different commercially available ACC systems remain to be untangled. Therefore, to fill this gap, this study investigates the car-following characteristics of various ACC systems and compares them with human drivers based on the open-access OpenACC database. We first identify the proper surrogate car-following model for denoting the driving behaviors of ACC vehicles and HDVs from five widely used car-following models, among which the best-fitted one is the intelligent driver model. Then, we implement the Gaussian mixture model and Jensen-Shannon divergence to describe the similarities between ACC systems and human drivers. Moreover, the string stability of different ACC platoons in various traffic conditions are investigated with a series of simulation experiments. Results show that all the ACC systems are string unstable, and more unstable than human drivers. The behavior of the Ford-ACC system is most similar to human drivers with a relative low instability, while the Peugeot-ACC system behaves most differently to human drivers, and aggressively with the highest instability.

No abstract

No abstract