Template matching is a popular method for visual recognition. However the method has two major problems: (1) Much time is required to detect the target object, and (2) Mistakes in recognition of the target object occur is some cases. In order to solve problem, (1) we apply a fuzzy neural network algorithm. However, problem (2) remains to be solved. In this paper, we propose a robust visual recognition algorithm for autonomous mobile robot navigation. The proposed algorithm employs Fuzzy Template Matching with a target object as well as similar objects in the environment. As a result, the robot verifies its tentative recognition by comparing the likelihood of plus elements (target objects) with that of minus elements (similar objects). The validity of the proposed navigation system is shown through some experimental results.