This paper describes an automated inspection robot for detecting tile exfoliation and a new diagnostic method for determining its existence and extent. The robot moves quickly along a vertical wall and stops to detect a tile's inner condition using a hammering sound. Tile separation commonly comprises outer exfoliation where the tile separates from the mortar concrete and inner exfoliation where the space between the substrate and the mortar concrete deteriorates. In order to detect these two exfoliations, we focused attention on wavelet analysis, which enables us to analyze the frequency element of the sound waveform on time phase continuously. By comparing the wavelet volume rate expressing the characteristics of tile deterioration for several scale tiles and striking hammers, the quantitative detection and its scale effects of visually distinguishing the two exfoliation modes was established. The automated robot and the diagnostics method were used to perform a fast and highly accurate inspection of the outer tile wall.