Loading...
Publication
Development of a 3D printed prosthetic hand and forearm controlled by Arduino
| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 4.82 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
A mão é uma ferramenta de elevada complexidade e a sua perda leva à total alteração
de vida de um individuo. A perda desde membro faz com que a pessoa deixe de conseguir
realizar as tarefas mais simples do quotidiano o que leva uma perda de qualidade de vida. A
utilização de uma prótese é a solução proposta para este problema. Este trabalho tem como
objetivo o desenvolvimento de uma mão protética fabricada por impressão 3D e controlada
através de uma placa Arduíno. Tentando assim criar um produto acessível e efetivo.
O dispositivo foi desenhado utilizado o software Fusion 360 e impresso em PLA utilizando
duas impressoras 3D, Prusa Mini Plus e Anycubic Kobra 2 Pro. A flexão e extensão dos cinco
dedos é realizado por motores servo MG995 fixos no antebraço, que puxam fios presos na
ponta de cada um. A abdução e adução do polegar é realizada por um motor servo SG90
embutido na mão. Os atuadores são controlados por joysticks, um para a rotação do
antebraço e outro para flexão/extensão dos dedos e abdução/adução do polegar. Foram
adicionados um sensor de temperatura e um de pressão, na ponta de cada dedo. Os sensores
de pressão permitem que o sistema limite a força aplicada num objeto agarrado pela mão. Os
sensores de temperatura permitem dar informação ao utilizador sobre a temperatura do objeto
que está a segurar. Todos estes foram encapsulados na pontas dos dedos com silicone. O
protótipo conseguiu realizar cinco gestos pré-programados e foi capaz de agarrar diversos
objetos utlizados no quotidiano. Mostrou que o sistema de atuação necessita de ser otimizado,
pois uso de cabos dificulta a realização de movimentos precisos. Grande parte da força
aplicada pelos motores é utilizada para vencer a resistência destes cabos e o atrito entre
componentes, reduzindo a eficiência do produto. A última etapa do trabalho está relacionada
com o desenvolvimento de uma interface pessoa-máquina baseada no reconhecimento de
gestos. Com base num conjunto de gestos pré-programados definidos, foi criado um corpus
para treino e teste dos algoritmos. Os dados recolhidos foram segmentados e posteriormente
foi feita a extração de características temperais e espectrais, nomeadamente RMS, VAR, STD
e PSD. A classificação foi realizada por uma rede neuronal e support vector machine (SVM)
tendo-se obtido uma precisão de 91,2% e 87,4% na classificação, respetivamente.
Conclui-se que mão prostética é capaz de realizar pelo menos cinco gesto préprogramado e de agarrar objetos do quotidiano. Apesar de se terem sido calibrados os
sensores não se encontram ativos no protótipo e ainda é necessário a calibração dos
sensores de pressão após a adição do silicone. Apesar dos resultados de AI esta componente
ainda necessita de mais estudo para ser viável.
The hand is a highly complex tool, and its loss can completely change an individual's life. The loss of this limb means that the person is no longer able to perform the simplest everyday tasks, which leads to a loss of quality of life. The use of a prosthesis is the proposed solution to this problem. The aim of this work is to develop a prosthetic hand manufactured using 3D printing and controlled using an Arduino board. The aim is to create an affordable and effective product. The device was designed using Fusion 360 software and printed in PLA using two 3D printers, Prusa Mini Plus and Anycubic Kobra 2 Pro. Flexion and extension of the five fingers is carried out by MG995 servo motors fixed to the forearm, which pull wires attached to the tip of each finger. Thumb abduction and adduction is performed by an SG90 servo motor built into the hand. The actuators are controlled by joysticks, one for forearm rotation and the other for finger flexion/extension and thumb abduction/adduction. A temperature sensor and a pressure sensor have been added to the tip of each finger. The pressure sensors allow the system to limit the force applied to an object gripped by the hand. The temperature sensors provide the user with information about the temperature of the object they are holding. All of these have been encapsulated in the fingertips with silicone. The last stage of the work is related to the development of a person-machine interface based on gesture recognition. Based on a set of pre-programmed gestures, a corpus was created for training and testing algorithms. The data collected was segmented and then temporal and spectral characteristics were extracted, namely RMS, VAR, STD and PSD. The classification tests were conducted using neural networks and support vector machines (SVM), with an accuracy of 91,2% and 87,4%, respectively. It was concluded that the prosthetic hand can perform at least five pre-programmed gestures and hold everyday objects. Despite being tested, functional and mounted on the prototype, the sensor system is not active and the encapsulated pressure sensors still need to be calibrated. AI showed satisfactory results with a NN of 91,2% and an SVM of 84,7% but this component needs further study to be viable.
The hand is a highly complex tool, and its loss can completely change an individual's life. The loss of this limb means that the person is no longer able to perform the simplest everyday tasks, which leads to a loss of quality of life. The use of a prosthesis is the proposed solution to this problem. The aim of this work is to develop a prosthetic hand manufactured using 3D printing and controlled using an Arduino board. The aim is to create an affordable and effective product. The device was designed using Fusion 360 software and printed in PLA using two 3D printers, Prusa Mini Plus and Anycubic Kobra 2 Pro. Flexion and extension of the five fingers is carried out by MG995 servo motors fixed to the forearm, which pull wires attached to the tip of each finger. Thumb abduction and adduction is performed by an SG90 servo motor built into the hand. The actuators are controlled by joysticks, one for forearm rotation and the other for finger flexion/extension and thumb abduction/adduction. A temperature sensor and a pressure sensor have been added to the tip of each finger. The pressure sensors allow the system to limit the force applied to an object gripped by the hand. The temperature sensors provide the user with information about the temperature of the object they are holding. All of these have been encapsulated in the fingertips with silicone. The last stage of the work is related to the development of a person-machine interface based on gesture recognition. Based on a set of pre-programmed gestures, a corpus was created for training and testing algorithms. The data collected was segmented and then temporal and spectral characteristics were extracted, namely RMS, VAR, STD and PSD. The classification tests were conducted using neural networks and support vector machines (SVM), with an accuracy of 91,2% and 87,4%, respectively. It was concluded that the prosthetic hand can perform at least five pre-programmed gestures and hold everyday objects. Despite being tested, functional and mounted on the prototype, the sensor system is not active and the encapsulated pressure sensors still need to be calibrated. AI showed satisfactory results with a NN of 91,2% and an SVM of 84,7% but this component needs further study to be viable.
Description
Keywords
Impressão 3D Arduíno Próteses Mioelétricas Machine Learning 3D printing Arduino Myoelectric prostheses Machine Learning