This project develops and simulates flexible, inkjet-printed memristors for neuromorphic computing and Echo State Networks (ESNs). Using low-cost inkjet printing with materials like graphene and hexagonal boron nitride, the memristors show the essential pinched hysteresis loop for synaptic computing. MATLAB and Simulink are used for empirical and physical modeling, and the memristors are integrated into a crossbar array for ESN implementation, improving data processing efficiency. This approach offers a scalable, energy-efficient solution for future neuromorphic systems.

The integration of machine learning (ML) into flexible electronics has significantly enhanced the processing and analysis of complex, large-scale data generated by sensors. Our research focuses on combining ML algorithms with inkjet-printed sensors to improve accuracy and overcome limitations associated with minimal fabrication processes. Various ML algorithms have been successfully used to analyze physiological signals collected by flexible sensors, driving advancements in health monitoring and other applications. The next step is to implement ML directly on hardware designed specifically for flexible electronics.

Deployment of machine learning models (ML) on resource-constrained devices, referred to as edge devices. These devices, which include microcontrollers and low-power sensors, can perform inference tasks without needing to connect to cloud services.  This innovation enables real-time processing and decision-making at the edge of the network, enhancing performance while minimizing resource consumption.