Closed-Loop Physical Reservoir Computing with Optogenetic Control in Biological Neural Networks
Jie Li, Yin Deng, Yarong Lin, Zeying Lu, Xiaojuan Sun, Yueheng Lan, Longze Sha, Lili Gui, Kun Xu · 2024
This study introduces physical reservoir computing (PRC) using in vitro biological neural networks for efficient signal processing and computation. The system employs optogenetic control and First-Order Reduced and Controlled Error (FORCE) learning to enable a virtual car to avoid obstacles. Online neuronal activity is captured with a CMOS camera based on calcium imaging and optical stimulation is performed via a digital micromirror device (DMD). Neuronal action potentials, induced by light, guide the movement of the car. The adaptive learning system demonstrates effective error minimization and robust obstacle avoidance, advancing neuro-engineering applications and intelligent biohybrid systems.