Deployment of robotic underwater vehicles (RUVs) in unpredictable underwater environments necessitates the development of an intelligent mission controller (IMC) that would facilitate autonomous navigation of predetermined waypoints while identifying objects, avoiding obstructions and building a model of what has been sensed in real time. The IMC should balance a preference for following a previously determined mission path and dynamically re-planning a mission in order to avoid obstacles and minimize pre-planned path deviation. This paper presents an IMC that combines an efficient three-dimensional world model representation with a frustum culling trajectory search method to create an intelligent, adaptive path planning algorithm balancing safe obstacle avoidance with minimal pre-planned path deviation. The IMC is tested in a real-world environment and results are discussed. Future improvements of this IMC will also factor in the ability to identify regions of interest off a mission path and explore them as necessary.