The main objective of this paper is to study and formalize the TakeOver Control in Collaborative Virtual Environments for Training (CVET). The TakeOver Control represents the transfer (the take over) of the interaction control of an object between two or more users. This paradigm is particularly useful for training scenarios, in which the interaction control could be continuously exchanged between the trainee and the trainer, e.g. the latter guiding and correcting the trainee's actions. The paper presents the formalization of the TakeOver Control followed by an illustration focusing in a use-case of collaborative maritime navigation. In the presented use-case, the trainee has to avoid an underwater obstacle with the help of a trainer who has additional information about the obstacle. The use-case allows to highlight the different elements a TakeOver Control situation should enforce, such as user's awareness. Different TakeOver Control techniques were provided and evaluated focusing on the transfer exchange mechanism and the visual feedback. The results show that participants preferred the TakeOver Control technique which maximized the user awareness.