Triboelectric flexible sensors (TFS) hold great potential for applications in human health monitoring and human-machine interaction (HMI), where both exceptional comfort and a wide linear sensing range are essential. However, their practical deployment is often hindered by inherent challenges, including limited flexibility, narrow linear response windows, and insufficient sensitivity to subtle pressure variations. In this work, MXene nanosheets and amino-functionalized carbon nanotubes (NH₂-CNTs) were strategically integrated into poly (vinylidene fluoride- co-hexafluoropropylene) (PVDF-HFP) and thermoplastic polyurethane (TPU), respectively, to enhance their triboelectric performance. By leveraging the coupling effect of electrostatic repulsion alongside solution surface tension during the electrospinning process, and employing ultrasonic dispersion technology, PVDF-HFP/MXene and TPU/NH₂-CNT composite membranes with multi-level micro-conical surfaces were fabricated as the negative and positive triboelectric layers. This enabled the construction of a spacer-free, asymmetric dual micro-cone TFS. The resulting TFS demonstrated outstanding performance, characterized by an ultrathin form factor, excellent breathability, a broad linear range (0.995 linearity from 1.25 Pa to 100 kPa), and an ultra-low limit of detection (1.25 Pa). These superior attributes have enabled the effective construction of a real-time human health monitoring system. Furthermore, integrating the TFS into a robotic manipulator and coupling them with a deep learning model facilitated the development of an intelligent gait and fruit recognition system, the accuracies reached 98.05 % and 98.3 %, respectively. The proposed self-powered, asymmetric dual micro-cone sensor and its associated system offer valuable insights for advancing human–machine interaction.