Autonomous Vehicle Control in CARLA Simulator Using Reinforcement Learning

Autonomous vehicles (AVs) have emerged as a transformative technology that promises safer and more efficient transportation systems. Central to the development of autonomous driving is the ability to make real-time decisions in complex and dynamic environments. Reinforcement learning (RL) has gained considerable attention as a promising approach that enables AVs to learn control policies directly from interaction with their environment. In this paper, two RL algorithms for autonomous driving were developed in the CARLA autonomous driving simulator: one based on the Deep Q-Learning (DQN) algorithm and the other based on the Deep Deterministic Policy Gradient (DDPG) algorithm. The evaluation of the developed algorithms was carried out using the user-defined metrics and shows promising results in learning from interaction.