The development of the human-machine interface (HMI) is endeavored to find effective approaches to interact with machines by applying emerging technologies. Triboelectric nanogenerator (TENG) can convert mechanical stimuli to electricity, which not only shows great potential in sensing but also is widely used in various HMI applications. This paper proposed a TENGbased hexagonfractal touchpad (HTPad) system using two channels to realize 18 sliding patterns from 3 different modes and a signal recognition module. A onedimensional convolution neural network (1D CNN) model is proposed for the recognition of the sliding direction signal with 96.5% accuracy, and handwriting digit signals collected by the touchpad can be recognized with a modified model with 99% accuracy. The proposed TENGbased hexagonfractal touchpad is easy to fabricate, scalable, and with high sensitivity. Furthermore, the recognition model can serve as a unified platform for different recog.nition tasks with little computational cost, which reveals great potential in HMI applications.