Underwater vehicles are increasing their autonomous capabilities more and more in order to carry out more complex and longer missions. This basically requires operational resilience and efficient energy consumption to succeed in persistent presence. Some of current Autonomous Underwater Vehicles (AUVs) have a large degree of self-governance but most of them lack self-management (e. g. auto-maintenance before, during, and after missions). This paper introduces the autonomic computing concept to AUV control architectures in order to explore a solution that endows AUVs with resilience and greenness. One of the attractive characteristics of self-managed AUVs is that automatic functions from self-managing capabilities are executed in background, i. e. the deliberative control layer gets rid of tasks that are now placed in the reactive one. This paper also presents a review of the approaches for self-managed systems, a discussion on suitability of current autonomic technologies, and future research directions.