This study introduces an approach to navigate dynamic environments through the fusion of Long Short-Term Memory (LSTM) networks, Dijkstra’s algorithm, and Lyapunov stability theory. This integrated method capitalizes on the strengths of each component: LSTM’s capability for predicting dynamic obstacles, Dijkstra’s algorithm for efficient pathfinding, and Lyapunov’s theory for ensuring navigational stability. By synergizing these elements, the approach provides a robust solution for autonomous systems to navigate unpredictable terrains with enhanced safety, reliability, and efficiency. Through theoretical analysis and empirical evaluation, this research demonstrates the effectiveness of the integrated framework in improving the autonomy, reliability, and safety of navigation systems. The findings highlight the potential of combining machine learning, algorithmic pathfinding, and control theory to enhance the navigational strategies of autonomous vehicles and robots, particularly in unpredictable environments.