<p>Tactile sensors are one of the core components for intelligent robots to realize human-like tactile sensing functions. For the application scenarios of material classification and roughness recognition, an enhanced flexible triboelectric tactile sensor (FTTS) is proposed in this study. The FTTS is prepared based on a single-electrode triboelectric nanogenerator (SE-TENG), which generates the corresponding electrical signals by coming in contact with objects. To enhance the properties of the SE-TENG, the synergistic enhancement effect on the electrical output of the SE-TENG was explored by controlling the addition ratio of MXene and Carboxylated cellulose nanofibers (CNF-C). The charge-trapping mechanism of the charge-trapping layer was analyzed, and the effects of the type and thickness of the charge-trapping layer on the performance of the FTTS were systematically investigated. Inexpensive and simple fabricated braided microstructures were prepared using the template method, which showed the optimum capability in terms of self-cleaning and electrical output performance. With these improvements, the FTTS showed a voltage sensitivity of 2.88 V/10⁴ Pa in the pressure response range of 10–400 kPa, a linearity of 0.993, and a linear trend in the fitting curve. At the same time, the FTTS has a stable response frequency, fast response time, and extreme cycling stability (over 10000 cycles). In addition, in terms of roughness recognition, the FTTS can accurately recognize samples with different surface roughness. In terms of material classification, the accuracy of classifying 11 materials with the visual geometry group (VGG) network reaches 96.08%. Based on these findings, FTTS-based tactile sensors offer diverse options in the future direction of electronic skin and tactile sensing.</p>