AI-Assisted Electronic Digital Twin for Capacitive Pressure Sensors Controlling Robotic Gripping

The paper focuses on the analysis of the Artificial Intelligence (AI) classification of the pressure forces to control the gripping of a robotic hand. The goal is to structure an Electronic Digital Twin (EDT) model of capacitance sensors to classify robotic finger movements. The model simulates a robotic application of two fingers gripper (Figure (a)) by means of a circuit model based on the capacitive effect principle. The proposed approach is able to classify and apply a control signal using an AI engine trained by circuit simulations (Figure (b)) providing the theoretical hand responses to pressure forces. The AI analysis is initially performed to select the best algorithm for the classification of the gripping and its control (Figure (c)) by processing a dataset of an electronic glove characterized by different time series signals of capacitance sensors. Compared with other AI supervised algorithms, such as Random Forest (RF) and Artificial Neural Network (ANN), the Convolutional Neural Networks (CNNs) exhibits the best performance for complex capacitive sensor systems. The CNN-EDT model is then successively applied to a system of two coupled capacitive sensors (Figure (d)) by proving the possibility to process even more complex robotic control systems, such as an electronic glove. The EDT is constructed by using LTSpice and KNIME open-source tools.