Boosting Artificial Olfaction: Visual Cues-Enhanced Gas Classification by a Bimodal Neuromorphic Device

Area of Science:

• Neuromorphic Engineering
• Artificial Olfaction
• Bio-inspired Sensors

Background:

• Artificial olfactory sensors face challenges in balancing accuracy, size, and power consumption.
• Large sensor arrays improve accuracy but increase system size and power demands.
• Biological multisensory synergy offers a potential solution for overcoming these limitations.

Purpose of the Study:

• To develop a miniaturized, low-power visual-olfactory bimodal neuromorphic device.
• To emulate biological perceptual fusion for enhanced gas recognition.
• To overcome the trade-offs inherent in current artificial olfactory systems.

Main Methods:

• Proposed a visual-olfactory bimodal neuromorphic device inspired by biological multisensory synergy.
• Emulated biological perceptual fusion, including bimodal perceptual weighting and enhancement.
• Utilized a small active area (148 µm²), low static power consumption (3.4 µW), and low operating voltage (1 V).

Main Results:

• Achieved ppb-level sensing performance for gas classification and concentration identification.
• Demonstrated a gas classification accuracy of 98.27% with the bimodal strategy, significantly outperforming the unimodal mode (52.24%).
• Enabled precise discrimination of mixed gases with overlapping sensing signatures.

Conclusions:

• The proposed bimodal perception strategy provides a unit architecture for miniaturized, low-power, and highly accurate artificial olfactory systems.
• This bio-inspired approach paves the way for next-generation multimodal neuromorphic sensing.
• The device overcomes limitations of existing artificial olfactory sensors through integrated visual and olfactory inputs.