Recently, there has been an increasing interest for applying techniques of machine learning in autonomous agent learning. The autonomous ability of agents to detect and adapt to their environment enhances their adaptability and efficacy in completing numerous activities. A learning framework known as Reinforcement learning (RL) learns an agent how to act in a way that maximizes the value of a reward signal. Within the domains of agentics, agents trained to perform various tasks through trial and error, by employing reinforcement learning. In this paper, by utilizing one of the reinforcement learning algorithms, Deep Q-learning Network (DQN). DQN, used for autonomous agent learning, the agent can learn and determine the optimum policy that will guide it to its destination from its interactions with the environment. The agent's decision-making ability allows it to modify its policy as a response to observable circumstances. Experiments performed in a simulated continuous environment, the revealed results that presented the ability of an agent to react in a response to changing conditions and demonstrated that the agent was effectively adapting its behavior in accordance with the learned policy.
Rabah et al. (Tue,) studied this question.