Wearable sensor has attracted a broad interesting in application prospect of human-machine interaction (HMI). However, most of sensors are assembled in the shape of gloves to accurately capture complex hand motion information, thereby seriously blocking the hand to complete complex tasks. Herein, a wearable pressure sensor based on drum-structured triboelectric nanogenerator (DS-TENG) is developed to capture subtle pressure signals for physiological signal detection, information encoding, gesture recognition, and wireless real-time robot control. The DS-TENG enables a limit of detection down to 3.9 Pa pressure, which can sensitively capture human micromotion signals of pulse, throat sounds, and wrist muscles contraction. Especially, combined with a microprocessor and Morse code, the DS-TENG worn on wrist can detect single-finger motion signals to translate into regular voltage signals, which is employed to encode information of 26 letters and subsequently decode into the corresponding letters. Furthermore, with an aid of machine learning, the DS-TENG array (2 × 2) can successfully achieve gesture recognition with high accuracy of 92% to wirelessly perform real-time robot control. Consequently, the wearable pressure sensor based on DS-TENG can capture subtle pressure signals for information encoding and wireless real-time robot control, which demonstrates extreme potential in the field of HMI and artificial intelligence.