Upravljanje kvadrokopterja z vmesnikom mišice-stroj

magistrsko delo

Matej Kramberger (Author), Aleš Holobar (Mentor)

Abstract

V delu smo zasnovali sistem za upravljanje kvadrokopterja v realnem času z uporabo vmesnika mišice-stroj. V programskem jeziku C# smo za operacijski sistem Windows izdelali aplikacijo, v kateri smo uporabili različne klasifikacijske algoritme iz odprtokodne knjižice Accord.NET. Klasifikacijo smo izvajali na računalniku s procesorjem Intel Core i7 2,8GHz ter 24 GB pomnilnika. Signale EMG smo zajeli s komercialno dostopno zapestnico Myo, ki omogoča zajem površinskih signalov EMG s podlahti. Uspešnost klasifikacije smo preizkusili na modelu kvadrokopterja Eachine E010, ki ga smo krmilili preko vmesnika nRF24L01 in mikrokontrolerja Atmel ATmega32u4 na razvojni plošči Arduino Micro. Klasificirane gibe smo uporabili za krmiljenje treh prostorskih stopenj kvadrokopterja. Giba ekstenzija in fleksija smo uporabili za nadzor naklona, pronacijo in supinacijo za nadzor nagiba ter ulnarno in radialno deviacijo za nadzor odklona. Za nadzor moči motorjev smo uporabili podatke inercijske merilne enote. Najboljše rezultate klasifikacije sta dajala algoritma SVM in k-NN, ki sta klasificirala s 95% pravilnostjo.

Keywords

elektromiogrami;kvadrokopterji;vmesniki mišice-stroj;zapestnica Myo;magistrske naloge;

Data

Language:	Slovenian
Year of publishing:	2018
Typology:	2.09 - Master's Thesis
Organization:	UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher:	M. Kramberger
UDC:	[004.9:004.5]:629.735(043.2)
COBISS:	21353238
Views:	911
Downloads:	106
Average score:	0 (0 votes)
Metadata:

Other data

Secondary language:	English
Secondary title:	Quadcopter Control with Muscle-Machine Interface
Secondary abstract:	We have designed a system for real-time quadcopter control by using the muscle-machine interface. In programming language C#, we have developed a Windows desktop application in which we have used different classification algorithms from the open-source library Accord.NET. Classification was conducted on the computer with Intel Core i7 2.8 GHz processor and 24 GB of memory. EMG signals were captured by commercial Myo armband, that supports acquisition of surface EMG signals from the forearm. We tested the accuracy of classification on quadcopter model Eachine E010, which we controlled via nRF24L01 interface and Atmel ATmega32u4 microcontroller on the Arduino Micro development board. We used classified movements to control three spatial degrees of freedom of quadrocopter. Wrist extensions and flexions were used for controlling pitch, pronation and supination for controlling roll and ulnar and radial deviation for controlling yaw. We used the inertial measurement unit data to control engine thrust. Best classification results were obtained by SVM and k-NN algorithms, with accuracy rate of 95%.
Secondary keywords:	electromyogram;quadrocopter;muscle-machine interface;Myo armband;
URN:	URN:SI:UM:
Type (COBISS):	Master's thesis/paper
Thesis comment:	Univ. v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko, Računalništvo in informacijske tehnologije
Pages:	X, 71 str.
ID:	10907812