International Conference on Computer and Knowledge Engineering

Home / 15th International Conference on Computer and Knowledge Engineering

DRL-based Decision-Making for Autonomous Vehicle Collision Avoidance

Authors :

Hoda Gholamrezaee¹ Seyedreza Taghizadeh² Ali Honarjoo³

1- Shiraz university 2- Shiraz university 3- Shiraz university

Keywords :

Autonomous vehicles،Collision risk،DRL،Internet of Vehicles

Abstract :

Internet of Vehicles (IoV) is a novel technology that plays a key role in realizing the dream of intelligent transportation systems (ITSs). By integrating diverse technologies, IoV enables information and data exchange between vehicles, infrastructure, and other devices, paving the way for innovative and practical services in the ITSs domain. On the other hand, the rise of autonomous vehicles (AVs) and smart car networks powered by the Internet of Things (IoT) offers immense potential in accelerating the development of ITSs. One of the most important challenges ahead is to ensure the safe movement of the AVs along a route, and it is necessary to evaluate the safety performance of the AVs during driving, because the risk of collision can be avoided or reduced by evaluating the driving safety. Therefore, in this article, an IoV-based network is introduced, in which AVs communicate with each other through road side units (RSU) in order to reduce the risk of the overall collision of the network. In the proposed model, a strategy has been developed to decide the efficient movement of AVs with the aim of avoiding collisions with obstacles or collisions with the least amount of damage for all vehicles in the network. To this end, this article explores a deep reinforcement learning (DRL) technique for enabling intelligent behavior decisions within the context of the discussed problem. The results presented in the article show the effectiveness of the proposed algorithm in reducing the amount of damage.