Exploring the Effectiveness of Deep Reinforcement Learning for Autonomous Robot Navigation

Due to the robot's requirement to make decisions in a dynamic and complicated environment, autonomous robot navigation is a difficult challenge. Positive outcomes have been achieved when using reinforcement learning (RL) to instruct robots in navigation. However, standard RL algorithms struggle with complex decision-making tasks and high-dimensional sensor inputs. Due to its ability to handle complex decision-making issues and learn directly from high-dimensional inputs, deep reinforcement learning (DRL) has emerged as a solution to these constraints. This paper provides an overview of the relevant literature and experimental findings to evaluate DRL's potential for use in autonomous robot navigation. This article delves into the application of DRL to various robotic environments, such as those on land, in the air, and underwater. This study summarizes previous research on the topic, exploring DRL algorithms such as DQN, DPG, and DRQN for their usefulness in autonomous navigation. The experiment's findings show that DRL can successfully navigate autonomously, even in complex surroundings. Three-dimensional mapping, goal achievement, map exploration, obstacle avoidance, and target tracking are some of the situations tested in this study. According to the article, DRL is a better option for autonomous navigation since it can better handle high-dimensional sensor inputs and difficult decision-making tasks than traditional RL algorithms. The development of novel algorithms, the integration of several modalities, and the investigation of practical applications are all highlighted as potential future paths for DRL research in autonomous robot navigation.