Multifunctional human visual pathway-replicated hardware based on 2D materials.

Area of Science:

• Artificial Intelligence
• Materials Science
• Neuroscience

Background:

• Current artificial vision hardware, while advanced, fails to fully emulate complex neural circuits.
• Replicating human visual pathways is key to unlocking more sophisticated AI vision capabilities.

Purpose of the Study:

• To propose and demonstrate a novel hardware system that replicates the human visual pathway.
• To integrate retina and visual cortex functionalities using 2D materials for advanced AI vision.

Main Methods:

• Developed crossbar arrays with split floating gate 2D tungsten diselenide (WSe2) unit devices to simulate retina and visual cortex.
• Designed peripheral circuits to replicate neural connectomics between simulated retina and visual cortex.

Main Results:

• Experimental demonstration of advanced multi-functions including red-green color-blindness processing.
• Achieved low-power shape recognition and self-driven motion tracking capabilities.
• Successfully replicated key aspects of human visual pathway neural circuits in hardware.

Conclusions:

• The proposed human visual pathway-replicated hardware advances artificial intelligence vision.
• This technology has significant implications for machine vision, driverless technology, brain-computer interfaces, and intelligent robotics.