A groundbreaking study on optoelectronic reservoir computing technology, led by Wan Changjin from Nanjing University, has recently been published in the prestigious journal Nature Communications. The research team has developed an innovative optoelectronic reservoir computing technology specifically tailored for video motion processing. Leveraging this technology alongside oxide functional devices, they have successfully created a biomimetic dynamic visual processing system dubbed Alpho-RC. This system prominently features an IGZO optoelectronic synaptic transistor chip serving as the reservoir layer, which facilitates high-dimensional nonlinear mapping. Meanwhile, a TaOx-based memristor chip with a 1T1R structure functions as the output layer, responsible for label inference through linear regression operations.

During rigorous testing, the Alpho-RC system demonstrated impressive recognition accuracy rates surpassing 90% across multiple standard human action datasets, underscoring its high performance capabilities. Furthermore, the system boasts significant energy efficiency, making it an ideal fit for resource-constrained environments such as edge devices and mobile platforms. The publication of this research signifies a notable milestone for Nanjing University in the realm of optoelectronic reservoir computing.