In this paper, we present an autonomous in-pipe robot navigating system using vision-based landmark recognition. We propose to use special features of the pipelines (such as elbows or branches) as landmarks by recognizing the shadows of the elbows and branches. To obtain consistent shadow images, the in-pipe robot equips with specially designed illuminator. By analyzing the shadow, the robot can easily identify these landmarks, detect the direction of the passage, and adaptively traverse through while continuously updating the map. The effectiveness of the proposed method is verified by real experiments using the in-pipe robot MRINSPECT V for inspecting inside of the miniature urban 8-inch gas pipeline structure.