Abstract: Based on studying the basic law of the Chinese natural language representation of path(NLRP),a visual navigation algorithm based on constrained natural language for indoor environments is designed through establishing the imagery map for mobile robot navigation.Firstly,landmarks are extracted from constrained Chinese text,relative directions and distances among the landmarks are obtained,and then those landmarks are mapped onto the imagery map.Secondly, key boot points,by which the path is divided into several segments,are extracted to realize the path planning of the mobile robot.For each segment,the matching degree of SURF(Speeded-Up Robust Feature) feature points between the referenced object's real-time pictures and sample pictures in the database is calculated,and the data of odometer are fused to ascertain the approximate location of the mobile robot.Then,based on the approximate location obtained in the previous segment, the running direction in the next segment is calculated,and the robot will run along it.The calculation and movement go analogically till the last segment.Experiment results demonstrate the feasibility,validity and robustness of the method.