INTELLIGENT SMART GLOVE SYSTEM FOR ROBOTIC HAND FINGER MOVEMENT PREDICTION

Abstract

Robot-assisted surgeries have advanced significantly, enabling complex procedures with minimal invasiveness and enhanced flexibility. This research proposes a machine learning (ML)–based approach for predicting finger movements of a robotic hand using a Smart Glove equipped with Light Dependent Resistor (LDR) sensors. The system employs an ESP-WROOM-32 microcontroller, integrated with Arduino IDE and Jupyter, to capture and refine real-time finger motions, including flexion and extension. These movements are mapped to the robotic hand through real-time integration. The dataset generated represents multi-output regression challenges, as multiple finger movements must be predicted synchronously. To address this, a K-Nearest Neighbors (KNN) regressor was applied, leveraging its ability to handle multi-output regression effectively. Model performance was evaluated using Root Mean Square Prediction Error (RMSPE), and real-time implementation demonstrated that the robotic hand accurately replicated finger motions corresponding to the Smart Glove inputs. The proposed method enhances control precision, reduces latency, and improves overall user interaction, offering potential applications in advanced prosthetics, artificial limb control, and remote robotic operations.