Publication

Abstract : Prosthetic arms are an effective way to enhance the quality of life for patients with upper-limb amputations. Though there are plenty of prosthetic arms commercially available on the market, functional prosthetic hands are still unaffordable for many due to complex control and design costs. In order to tackle this challenge, this paper explores the use of a swappable end-effector based prosthetic forearm with a machine learning-based advanced control unit. In this paper, two different end-effectors, such as a simple four-bar linkage based gripper and an anthropomorphic hand, have been implemented to test the prosthetic hand functionality using quadratic SVM based classification and a dual-stage control unit. The control unit of the proposed system consists of two levels of controls: first raspberry pi-based EMG data processing and classification, and then an end-effector microcontroller for the end-effector control, which helps in choosing two different types of end-effectors for the user. The design and dataset details are explained in this paper.