A Low-Cost Lightweight Prosthetic Arm with Soft Gripping Fingers Controlled Using CNN

Abstract

In this paper, a low-cost prosthetic arm design approach is introduced through a non-invasive technique for amputees. The arm is controlled by brain signals using electroencephalography (EEG) data obtained from an EEG headset. The prosthetic arm is fabricated from two different materials: a soft elastomer material for fingers which provides the arm with a high power-to-weight ratio and PLA 3D printing material for the limb. The driving force is a pneumatic circuit that helps to avoid the vibrations and noise induced by the servo motors in the currently existing commercial prosthetic limbs. A commercial headband (OpenBCI) was used to capture brain signals from the amputee's motor cortex via the headset's GUI and controlled using Raspberry Pi 4. Subsequently, two different learning algorithms were adopted, Support Vector Machine (SVM) and Convolutional Neural Network (CNN) to classify brain signals and transmit them to the prosthetic arm. The fingers are designed from elastomer material with a length of 70 mm and a wall thickness of 2 mm to increase the strength of the actuator and reduce the applied air pressure. A finite element analysis using ABAQUS software is conducted to investigate the stress on the soft-gripping fingers. A bending angle of 90 degrees was achieved while applying a pressure of 3.3 Pa.