Supervised training of voting automata for the surgeon's motion recognition during laparoscope surgery

Supervised learning of voting automata for the surgeon's right hand motion recognition constitutes the main result reported in the present paper. Within the framework of the project, aiming the design of scrub nurse robot a number of methods for recognizing the current stage of the surgery has been developed. Obviously no one of the methods separately can guarantee hundred percent correct recognition. Therefore, the voting automaton is employed to choose the best result. In this paper main attention is paid to the design of such voting automata using supervised learning techniques.