Over-actuated, hover-capable autonomous underwater vehicles (AUVs) are hybrid vehicles designed to serve a wide range of operations, from detailed survey tasks in a hover style to long-range exploration tasks in a flight style. Due to the over-actuated design and the different operating styles, it is challenging to operate these types of AUVs. This research developed a navigation system for over-actuated, hover-capable AUVs, with a primary focus on interconnections between guidance, control and path planning systems. A detailed 6-degrees of freedom mathematical model was developed to represent dynamics of these hybrid AUVs. The model was validated with horizontal plane manoeuvring trials carried out on the over-actuated, hover-capable Delphin2 AUV. The results showed that the model can satisfactorily predict the AUV's response over a range of operating conditions. Using this dynamics model, the guidance and control systems were designed to effectively and efficiently operate over a range of forward speeds with a seamless transition between hover-style and flight-style operation. A path planning system was tested, seeking for the collision-free path between two locations that requires the least energy for an AUV to navigate along. The proposed navigation system has been verified on the Delphin2 AUV through experiments. The comprehensive results have shown that the system can maintain excellent performance regardless of a range of forward speeds. By applying the proposed navigation system, it is also possible to launch a long-endurance AUV from a shore. The vehicle would be able to automatically navigate along an energy efficient path to perform, for instance, a seabed survey mission at a remote part of the ocean, then come back to the recovery point. This concept eliminates a ship from the AUV launching process, hence, making the operation more cost-effective.