نوع مقاله : مقاله پژوهشی
نویسندگان
دانشکده مهندسی مکانیک، دانشگاه صنعتی خواجه نصیرالدین طوسی، تهران، ایران
چکیده
کلیدواژهها
موضوعات
عنوان مقاله [English]
نویسندگان [English]
A self-driving car, known as an autonomous car (AC),, is a car that is capable of traveling without human input. Self-driving cars are responsible for perceiving the environment, monitoring important systems, and control, including navigation.
Also, automated driving is a new technology to reduce traffic accidents and improve driving efficiency. In this study, a deep reinforcement learning (DRL)-based decision-making policy for self-driving cars for Highway overtaking scenario is presented. In fact, first, a highway traffic environment is created where the target is to pass the agent through the surrounding vehicles with an efficient and safe maneuver. Then,, a hierarchical control framework is provided for the control of these vehicles, with high-level orders managing driving decisions and low-level orders monitoring the speed and acceleration of the vehicle. Then, a specific method based on deep reinforcement learning (DRL) called the deep deterministic policy gradient algorithm(DDPG) is used and evaluated to extract the decision-making strategy on the highway. The performance of the deep deterministic policy gradient (DDPG) algorithm has then been compared to the deep Q network algorithm (DQN) and results will be evaluated and examined. The simulation results show that the deep deterministic policy gradient (DDPG) method performs highway driving tasks efficiently and safely.
کلیدواژهها [English]