Led by Professor Sejoon Lee, the team designed the device to replicate how mechanoreceptors in human skin convert physical deformation into synaptic signals. The system utilizes two triboelectric nanogenerators (TENGs) coupled with a graphene-channel ion-gel-gated transistor. One TENG manages pre-synaptic spikes while the other handles post-synaptic output, allowing the circuit to harvest energy directly from movement or vibration.

Testing revealed the platform can reproduce hierarchical memory states, including sensory memory with 70-millisecond decay and short-term memory lasting up to 0.45 seconds. With repeated stimulation, these signals transition into long-term memory exceeding 2 seconds. The device also demonstrates spike-rate-dependent plasticity, a key learning mechanism that remains stable even when the material is bent.

In practical applications, the researchers integrated the technology into a single-layer artificial neural network to monitor human movement. The system successfully classified activities such as walking, sitting, and standing with 88.05% accuracy. Even under high-noise conditions, the platform maintained 75% performance, suggesting a viable path for battery-free health monitoring, smart prosthetics, and advanced human-machine interfaces.