Fig. 7: Neuromorphic robotics systems.
From: Advanced AI computing enabled by 2D material-based neuromorphic devices
a Schematic illustration of the Au/MoS2/Ag threshold switching memristors (TSM) device (top) and an optical image of the fabricated device (bottom)58. b Output current of the artificial nociceptor for increasing pulse amplitudes (0.6 V to 1.2 V) at a fixed pulse width of 1 ms (top) and output current of the artificial nociceptor for varying pulse widths (10 µs to 2 ms) at a constant pulse amplitude of 1 V (bottom). c Schematic of biological nociceptor and its key features (top) and block diagram illustrating the architecture of the MoS2 TFT-based artificial nociceptor (bottom). d Schematic illustration of the Ag/SnSe/Au TS device (top) and cross-sectional HRTEM image of the device (bottom)59. e Temporal evolution of the membrane potential (black) of a LIF neuron in response to a series of input spikes (green). The neuron fires when the membrane potential exceeds the threshold, generating an output spike (red). f Stochastic LIF neuron operation. g Classification accuracy of the TSM-based neural network on the MNIST dataset. h Schematic and optical image of the flexible dual-gate synaptic transistor60. i Confusion matrices for human activity recognition (left) and drone flight mode classification (right) using SNN.
