Advanced hardware and software systems can be employed to equip autonomous agents with the ability to explore their surroundings while minimizing the risk of collisions. Ensuring Artificial Intelligence (AI) safety is crucial to providing reliable service to consumers in various sectors, such as the military, education, healthcare, and automotive sectors. One way to increase the accuracy and performance of an AI agent is by guiding it in a specific environment through human intervention. This study presents the design of a system that can be controlled by two methods. The first method uses an AI controller with the Proximal Policy Optimizer (PPO) algorithm to guide the agent automatically. The second method requires human interaction to guide the agent using keyboard inputs (manual controller). The Machine Learning Agents Toolkit (ML-Agents) was used to train the agent with a PPO algorithm with an Intrinsic Curiosity Module (PPO + ICM). The objective of this study is to demonstrate the advantages and disadvantages of the two presented methods. We conducted both methods for ten minutes and measured the Goal (G), Collision (C), and GC-ratio or Goal (G) Collision (C) Ratio. The GC-ratio is a key metric for measuring the frequency of collisions between the agent and obstacles. The results show that the AI controlled method achieved 45 goals and 27 collisions, resulting in a GC-ratio of 0.66. On the other hand, the manually controlled method provided 30 goals and 6 collisions, resulting in a GC-ratio of 0.83. Each method has its own merits and limitations, and the decision to choose one over the other ultimately depends on the specific needs of the application.