Photo-Assisted Energy Harvesters and Sensors Using PVDF Matrix-Stabilized α-FAPbI ₃ Perovskite with AI/ML-Based Human Kinematic Detection

Recent advancements in energy-harvesting technologies have led to the development of systems that efficiently convert mechanical energy into electrical power, offering sustainable solutions for a wide range of applications. In particular, piezoelectric materials are paving the way for high-performance, flexible devices capable of energy conversion, storage, and sensing, addressing the growing demand for self-powered and eco-friendly systems. In this study, we present a simple approach to stabilize α-FAPbI 3 by incorporating it into a PVDF matrix, resulting in a stable composite film. Optimized FPC 3 composite containing 300 μL of α-FAPbI 3 /g in PVDF achieved high electroactive-phase formation (∼96%). A comparative study demonstrated that the black phase (α-FAPbI 3 ) outperformed the yellow phase (δ-FAPbI 3 ) due to superior interfacial interactions. The optimized device, FPNG 3, generated a peak-to-peak output voltage of ∼148.4 V and a maximum power density of ∼81.72 μW/cm 2 . Additionally, the device exhibited high pressure sensitivity (∼1.036 kPa/V at 10-Hz frequency and a moderate pressure range) with the ability to sense various biomechanical movements. The light-assisted nanogenerator device displayed output variations under no light, white light, and UV light exposure, making it suitable for photoassisted energy harvesting. A ∼63.5% increase in output voltage was observed from the photoassisted energy-harvesting device under white light. Overall, this research highlights the multifunctionality of the FAPbI 3 –PVDF composite in energy conversion, storage, and sensing, along with an effective method for stabilizing the metastable, photoactive α-FAPbI 3 . Additionally, a machine learning framework utilizing LSTM networks was employed for real-time gesture recognition and classification of the energy-harvesting behavior of the FPNG device. This approach enables real-time performance optimization, enhancing applications in sign language recognition, human–computer interaction, and sensor systems.