Closed-Loop Haptic–Thermal Perception with Memristor-Based Spiking Neurons for Embodied Neuromorphic Intelligence

TY - JOUR

T1 - Closed-Loop Haptic–Thermal Perception with Memristor-Based Spiking Neurons for Embodied Neuromorphic Intelligence

AU - Huang, Tianci

AU - Li, Haotian

AU - Dong, Zilong

AU - Yuan, Zuqing

AU - Hua, Qilin

AU - Hu, Weiguo

AU - Shen, Guozhen

N1 - Publisher Copyright: © 2025 Wiley-VCH GmbH.

PY - 2026/3/16

Y1 - 2026/3/16

N2 - Advances in embodied intelligence necessitate the integration of tactile and thermal sensing in artificial sensory systems to enable adaptive human–robot interactions, which compensate for insufficient or entirely unavailable visual information in contact-haptic operations. Here, a closed-loop haptic–thermal perception system featuring silver nanowire (AgNW) memristors with dual-mode pressure-temperature sensors is presented. Optimized via spin-coating AgNWs and ALD-grown Al2O3 encapsulation, AgNW memristors demonstrate bidirectional threshold switching behavior with ultralow leakage current (<1 nA), sub-1V threshold voltage, and ambient stability. Flexible dual-mode sensors convert external stimuli into electrical signals, mimicking the human skin's perception of pressure and temperature. Sensory stimuli are processed by AgNW memristor-based spiking neurons, which can fuse simulated information from dual-mode sensors into a spike sequence and classify via convolutional neural networks (CNNs), emulating four haptic–thermal perceptual levels in a robot hand—from gentle touch to extreme discomfort. This architecture enables energy-efficient, low-latency decision-making that facilitates artificial nociceptive reflexes for safe human–robot interaction while advancing neuromorphic devices for next-generation wearable electronics and embodied intelligence.

AB - Advances in embodied intelligence necessitate the integration of tactile and thermal sensing in artificial sensory systems to enable adaptive human–robot interactions, which compensate for insufficient or entirely unavailable visual information in contact-haptic operations. Here, a closed-loop haptic–thermal perception system featuring silver nanowire (AgNW) memristors with dual-mode pressure-temperature sensors is presented. Optimized via spin-coating AgNWs and ALD-grown Al2O3 encapsulation, AgNW memristors demonstrate bidirectional threshold switching behavior with ultralow leakage current (<1 nA), sub-1V threshold voltage, and ambient stability. Flexible dual-mode sensors convert external stimuli into electrical signals, mimicking the human skin's perception of pressure and temperature. Sensory stimuli are processed by AgNW memristor-based spiking neurons, which can fuse simulated information from dual-mode sensors into a spike sequence and classify via convolutional neural networks (CNNs), emulating four haptic–thermal perceptual levels in a robot hand—from gentle touch to extreme discomfort. This architecture enables energy-efficient, low-latency decision-making that facilitates artificial nociceptive reflexes for safe human–robot interaction while advancing neuromorphic devices for next-generation wearable electronics and embodied intelligence.

KW - AgNW

KW - artificial spiking neuron

KW - embodied intelligence

KW - haptic–thermal perception

KW - memristor

UR - https://www.scopus.com/pages/publications/105020767348

U2 - 10.1002/adfm.202523270

DO - 10.1002/adfm.202523270

M3 - Article

AN - SCOPUS:105020767348

SN - 1616-301X

VL - 36

JO - Advanced Functional Materials

JF - Advanced Functional Materials

IS - 22

M1 - e23270

ER -