Reinforcement Learning: Applications in Autonomous Systems and Robotics

ABSTRACT Reinforcement Learning (RL) has become a central approach in modern artificial intelligence, particularly in domains that require sequential decision-making under uncertainty. Unlike traditional machine learning techniques that rely on labeled datasets, RL enables an agent to learn optimal actions through interaction with an environment by maximizing cumulative rewards. This makes it especially suitable for autonomous systems and robotics, where environments are dynamic and often unpredictable. In recent years, the integration of Deep Learning with RL, commonly referred to as Deep Reinforcement Learning (DRL), has significantly improved the ability of robots to perceive, learn, and act in complex real-world scenarios. This review paper examines the role of RL in key areas such as robot navigation, manipulation, multi-agent coordination, human-robot interaction, and industrial automation. It also highlights the major challenges, including sample inefficiency, safety risks, and real-world deployment limitations. By analysing current developments and practical applications, this paper provides insights into the future direction of RL in autonomous systems and robotics.