The depth information of object surface topography is of great significance for intelligent robots to detect object surface features and human-computer interaction. Inspired by the mechanism of animal hair perception, a novel bionic magnetostrictive tactile sensor unit and array are designed. Based on the inverse magnetostrictive effect, Euler-Bernoulli beam theory, and Hooke′s law, the output voltage model of depth detection is derived. The simulation study determined the optimal bias magnetic field and array spacing. The output characteristics of the sensor unit under static and dynamic conditions were tested. In the depth detection range of 0. 05 ~ 4. 8 mm, the sensitivity is 185. 72 mV/ mm, and the response time and recovery time are 31 ms and 43 ms, respectively. It has good repeatability, and the coupling effect of the output voltage between the units in the sensor array does not exceed 2. 4% . The sensor unit and array are installed on the manipulator. When the appropriate sliding speed is selected to slide through the surface of different objects, various depths, and morphologies are accurately measured according to the output voltage waveform. The results show that the sensor unit and array can provide a reference for depth and morphology detection.