Autonomous Underwater Vehicles (AUVs) are integral to a growing number of applications, including oceanographic surveys, underwater structure inspection, and military operations. A significant challenge, however, is the autonomous navigation of these vehicles in complex environments characterized by both static and dynamic obstacles. This paper introduces a hybrid framework that integrates the A* pathfinding algorithm with the Support Vector Machine (SVM) to holistically address trajectory optimization and collision avoidance. Within this framework, the SVM processes sensor data (e.g., sonar) to classify the surrounding environment, constructing a high-accuracy map that distinguishes between navigable free space and obstacle zones. Subsequently, this map informs a cost function for the A* algorithm, which then computes an optimal trajectory from a starting point to a destination. Simulation results demonstrate that our proposed A*-SVM approach not only ensures effective collision avoidance but also exhibits superior performance in terms of path length and computational efficiency when compared to conventional methods.