Autonomous vehicle navigation can be divided into two major areas of research: Collision avoidance and Track-Keeping. This study focuses on Collision avoidance which is one of the major issues that unmanned autonomous vehicles have to face. Collision avoidance may be further grouped into classical and soft computing based categories. Classical techniques are based on mathematical models and algorithms, while soft-computing techniques are based on Artificial Intelligence. In this study, we focus on the Classical techniques and more specifically in the Potential Field Methods. The potential field algorithms rapidly gained popularity due to their simplicity and elegance. In other words, Potential Field Methods are generic, computationally efficient and generate naturally smooth trajectories. On the other hand, PFM algorithms experience local minima. Nevertheless, local minima for PFM are extensively studied in different environments; they have never studied in a Pure Dynamic Environment (PDE). PDE is a new dynamic environment in which all its elements are guaranteed to be dynamic at their initial state. In this way we have managed to identify and define the causes of Potential Field Agent local minima and trajectory inefficiencies in a number of collision scenarios within PDE. To efficiently and accurately identify and define these causes of local minima and traj ectory inefficiencies, we have introduced the novel concept of the Monovular Autonomous Agent Correlation. Based on this concept we have identified and mathematically defined the Trajectory Equilibrium State (TES) for the first time. This state is responsible for local minima and trajectory inefficiencies of Monovular Autonomous Agents in PDE. Because of TES identification and definition we have designed a lUle based mathematical algorithm that efficiently navigates the Autonomous Agents out of local minima and trajectory inefficiencies in PDE in a number of generic collision scenarios. The algorithm's performance is tested in a number of simulated water based collision scenarios.