In this paper we show how techniques that have been applied during the last years for autonomous manipulation in the air can be successfully adapted to a new challenging scenario: the underwater environment. Specifically, the developed techniques include among others: visual tracking of objects in the seabed, vision-based arm control and sensor-based grasping. Furthermore, a graphical user interface for specifying underwater manipulation actions and providing the corresponding task specification to the execution modules is also presented. This research is enclosed in the framework of our two ongoing projects in autonomous underwater intervention: RAUVI and TRIDENT.