Optoelectronic artificial synapse for lateral inhibition-enhanced retinal biomimicry

Area of Science:

• Neuroscience and Materials Science
• Optoelectronics and Artificial Intelligence

Background:

• Optoelectronic synaptic devices emulate visual neurons for in-sensor processing.
• Lateral inhibition enhances spatial selectivity and dynamic range in visual systems.
• Current systems lack efficient integration of lateral inhibition for complex tasks.

Purpose of the Study:

• To develop a novel optoelectronic artificial synapse incorporating lateral inhibition.
• To emulate neural mechanisms for advanced image processing.
• To enable efficient robotic control and machine vision systems.

Main Methods:

• Fabrication of a heterostructure device using In2O3, Cs2AgBiBr6 perovskite, and IGZO.
• Simulation of excitatory synaptic activity via light and lateral inhibition via electrical stimulation.
• Development of a lateral inhibition network for image recognition and integration with robotic arms.

Main Results:

• The novel heterostructure enhances optoelectronic response and synaptic plasticity.
• The device successfully replicates neural mechanisms like Mach bands and Hermann's grid.
• A lateral inhibition network achieved 97% image recognition accuracy, surpassing conventional networks (93%).

Conclusions:

• The developed optoelectronic synapse offers a cost-effective and energy-efficient solution for advanced machine vision.
• The device demonstrates potential for intelligent autonomous devices and bioinspired robots through seamless robotic integration.
• This work paves the way for sophisticated neuromorphic computing applications.