Abstract: First,the principles and application of vision sensors and tactile sensors used in endoscope navigation are introduced.Next,probability is assigned to the information of various sensors by analyzing the data from them.At last,Dempster-Shafer theory of evidence is used and various information of sensors is taken as evidence to navigate the endoscope.Experiment shows that the tiny change of single sensor can be filtrated and the remarkable change can be magnified with the multi-sensor information fusion algorithm for endoscope navigation,which is advantageous to the endoscope's insertion and can effectively solve the problem of endoscope perforation.