Bidirectional Photovoltage-Driven Oxide Transistors for Neuromorphic Visual Sensors

Area of Science:

• Neuromorphic Engineering
• Materials Science
• Computer Vision

Background:

• Biological vision relies on complementary photoexcitation and photoinhibition, which is difficult to emulate.
• Existing artificial vision systems face challenges in replicating the complexity of human visual perception.

Purpose of the Study:

• To develop a bidirectional photovoltage-driven neuromorphic visual sensor (BPNVS).
• To mimic the human visual system's ability to recognize colored and color-mixed patterns.
• To enable dynamic color recognition and object movement detection.

Main Methods:

• Monolithic integration of two perovskite solar cells (PSCs) with dual-gate ion-gel-gated oxide transistors.
• Utilizing PSCs as photoreceptors to convert visual stimuli into electrical signals.
• Employing oxide transistors for adjustable positive and negative photoresponses and neuromorphic signal generation.

Main Results:

• The BPNVS achieved 96% static color recognition accuracy using reservoir computing for feature extraction.
• Demonstrated the device's capability to mimic human vision for recognizing color and mixed patterns.
• Proposed a BPNVS mem-reservoir chip for object movement and dynamic color recognition.

Conclusions:

• The BPNVS represents a significant advancement in neuromorphic sensing.
• The developed sensor shows promise for complex pattern recognition and artificial vision applications.
• This work paves the way for more sophisticated bio-inspired visual systems.