Sub-1 ms Bio-Inspired Time-to-First Spike (TTFS) Neuromorphic Tactile System Using L-AFeFET and Threshold Switch Devices

초록

We propose a bio-inspired neuromorphic tactile system that achieves sub-1 ms time-to-first spike (TTFS) coding of pressure stimuli, bridging the latency and energy-efficiency gap between biological and artificial tactile systems. (i) A threshold switch (TS) and leaky anti-ferroelectric FET (L-AFeFET) neuromorphic devices are co-integrated to convert event-driven analog pressure inputs into TTFS-coded spikes within 1 ms, enabling real-time tactile encoding. (ii) To emulate biological signal processing mechanisms, the proposed system treats the entire array as a single receptive field, where spike timing is modulated by dynamically adjusting the L-AFeFET body bias. This approach suppresses redundant spiking and local interference, improving energy efficiency. As a result, the system achieves a 33.1% reduction in spike generation and a 52.8% reduction in synaptic operation compared to conventional TTFS coding, confirming enhanced energy efficiency of the bio-inspired lateral inhibition mechanism. This work establishes a pathway toward a high-efficiency neuromorphic tactile system, with strong potential for next-generation robotics and prosthetic applications.

키워드