Domestic researchers have developed next-generation intelligent sensors that detect external stimuli and learn and respond autonomously like human skin.
Professor Lee Nae-eung's research team at Sungkyunkwan University has developed an intelligent artificial tactile receptor array that mimics the function and structure of similar synapses, inspired by the human tactile perception system, and has implemented a new intelligent sensor platform based on this, they noted on the 29th. Tactile receptors detect external stimuli such as pressure, vibration, and temperature, converting them into action potentials and transmitting them to the brain.
Recently, the importance and role of artificial intelligence (AI) have been highlighted across various industries, particularly as physical AI, which implements and applies AI in actual physical environments, is emerging as a core foundational technology for autonomous systems in future industries. In physical AI, data input begins through sensors, prompting active research into intelligent sensor technologies that not only equipped with high-performance signal processing capabilities for efficient handling of sensor data but also mimic the mechanisms of the human somatosensory system.
The research team focused on the way human sensory organs initially process information, specifically the 'similar synapse structure' between sensory receptors and nerve endings. Inspired by Merkel cells, which slowly adapt to stimuli, and Pacinian cells, which quickly adapt, they developed a platform that integrates 16 sensory sensor units and synapse units (synaptic transistors) reflecting both adaptation characteristics.
The developed platform features a triboelectric sensor layer resembling human fingerprints and a synaptic transistor that remembers and reacts to stimuli, allowing it to recognize both slow and fast stimuli simultaneously.
Through experiments, it was confirmed that this sensor naturally changes and reacts according to the intensity, frequency, and form of mechanical stimuli by varying synaptic weights. Notably, even utilizing less than 10% of the total training data, it can recognize textures and surface patterns with over 90% accuracy, demonstrating significantly superior data processing efficiency compared to existing technologies.
The research team stated, "Sensors with AI functions embedded in their sensations are characterized by ultra-low voltage, ultra-low power, and high efficiency, presenting new technological possibilities in various fields such as intelligent robots, neuromorphic sensory systems, and wearable electronic skin," adding, "Particularly, as it can process external environmental data from the sensor stage, it is receiving attention as a key technology for implementing high-speed and high-efficiency autonomous AI systems in the future."
The research results were published in the international academic journal "Nature Materials" on the 28th.
References
Nature Materials (2025), DOI: https://doi.org/10.1038/s41563-025-02204-y