Fig. 4: AI-assisted active tactile perception strategy and validation.
From: A robust and omnidirectional-sensitive electronic antenna for tactile-induced perception
A Schematic of active tactile perception strategy. The E-Antenna is moved to interact with objects for gathering time-spatial features along the moving direction, a sliding window is used to integrate discrete tactile signals to construct a tactile sensing sequence, and a sequential neural network is designed to perceive environmental information and the antenna’s state. B Confusion matrix of the classification results of antenna’s states (S0-S5) where most classes achieve 100% classification accuracy. C Violin plot of the prediction error for five surfaces with varying curvatures (C1: 40°, C2: 61°, C3: 62°, C4: 63°, and C5: 65°), where the overall maximum error is less than 1o.
