A bee-inspired multisensory spiking neuron with Humid-Visual fused perception for authenticity recognition

Multisensory fusion allow artificial neurons to capture and process abundant signals for perceptual enhancement. Apart from the extensively studied visual-tactile fusion, reports on visual-humid perception remains a challenge, which plays a crucial role for living organism in activities such as foraging, weather detection and locating water source. Here we demonstrate a bee-inspired multisensory spiking neuron (MSN) that comprise a capacitive humidity synapse, a resistive photodetector and a threshold- switching memristor. The MSN encodes visual and humidity inputs into a sequence of spike events, mimicking the quintessential function of a bio-neuron, including action potential, threshold-driven spiking, refractory period, and intensity-modulated spike- frequency. The fused spikes can distinguish between humidity and visual signals by decoupling the output frequencies and amplitudes. Simulation results depict that the cross-modal sensing approach achieves a heightened accuracy (96.67%) in object recognition for verifying flower authenticity. This research opens up a new avenue for constructing humid-visual spiking neurons to advance multisensory perception and intelligent recognition capabilities.