Energy harvesting technologies, such as the triboelectric nanogenerator (TENG), which aims to scavenge high-entropy energy from the human body and living environment to power sensors, are becoming increasingly important in fields like healthcare, environmental monitoring, and wearable sensing. Since the collectable energies in living environment are typically irregular, scavenging them to power the widely distributed sensors while maintaining their inherent high accuracy requires complex power management with excessive power consumption, which makes it difficult even impossible for energy harvesting technology to be applied in many scenarios like wearable applications that generally require real-time sensing miniaturization, lightweight, and portability. To conquer this challenge, we proposed a calibration strategy (CS) that utilizes a shunt circuit to monitor the output of energy harvesting and calibrate the sensing signal, achieving real-time high-precision sensing. Our theoretical, computational, and experimental results demonstrate that CS enables TENG to function as a constant voltage source, capable of powering various commercial sensors (temperature sensors, opto-sensors, and humidity sensors) in real time. The sensing relative error achieved by CS can be as low as 0.87%, a level not previously attained in real-time powered sensors. As a proof of concept, we constructed a wearable multi-modal sensing system powered by the CS enhanced TENG, capable of real-time monitoring human motion (movement cadence, force on foot, surface temperature and relative humidity of skin) and environmental (light intensity, temperature, and relative humidity). This finding provides an effective strategy for energy harvesting technology to power sensors in real time while maintaining high accuracy, significantly advancing the practical application of energy-harvesting-based sensing systems.