Geri Dön

Knee joint pose estimation using deep neural network fortransfemoral prosthesis

Başlık çevirisi mevcut değil.

PDF İndir

  1. Tez No: 711898
  2. Yazar: YUNUS EMRE BADEM
  3. Danışmanlar: PROF. DR. YUKİO TAKEDA
  4. Tez Türü: Yüksek Lisans
  5. Konular: Makine Mühendisliği, Mechanical Engineering
  6. Anahtar Kelimeler: Belirtilmemiş.
  7. Yıl: 2020
  8. Dil: İngilizce
  9. Üniversite: Tokyo Institute of Technology
  10. Enstitü: Yurtdışı Enstitü
  11. Ana Bilim Dalı: Belirtilmemiş.
  12. Bilim Dalı: Belirtilmemiş.
  13. Sayfa Sayısı: 86

Özet

Özet yok.

Özet (Çeviri)

As the prosthetic technology progressed to robotics, a need for an online knee joint angle estimation algorithm for transfemoral prosthesis has risen. In order to estimate the knee joint angle, this thesis investigated the application of Long-Short Term Memory (LSTM) neural network that performs both as a high-level controller and low-level controller. Input is selected as the angular state of other residual joints which are namely, right hip joint, left hip joint and right knee joint. The angular state refers to a vector of data at any given state which includes angle, angular velocity, and angular acceleration of every residual joint in the lower limbs. Only the angular state on sagittal plane is used in this work. A dataset of 2899 gait recordings is expanded by means of digital signal processing. Moreover, stand-to-sit and sit-to-stand movements are procedurally generated. Six different deep LSTM neural networks are trained with a total of 105980 gait recordings and evaluated for varying gait speed, multiple movements, error rates and adaptability. Based on the results, a deep LSTM network can successfully respond to the intended movement. Moreover, the networks were able to successfully respond to varying walking speeds without any intervention. In addition, there is a strong correlation between angular state and the output knee joint angle that a neural network can extract. Based on the quantitative results, deeper networks should be preferred instead of wider networks in terms of accuracy and precision. The implemented network acted as both a high-level controller and a mid-level controller. The results suggest that deep LSTM networks is a desirable element in the control of active transfemoral prosthesis which provides a combined high and mid level controller.

Benzer Tezler

  1. Tez No

    342093

    Diz ve kalça eklem protezlerin değerlendirilmesi

    Evaluation of the infections of knee and hip joint prosthesis: A clinical experiment

    ABBAS TOKYAY

    Tıpta Uzmanlık

    Türkçe

    Türkçe

    2013

    Ortopedi ve TravmatolojiDokuz Eylül Üniversitesi

    Ortopedi ve Travmatoloji Ana Bilim Dalı

    PROF. DR. VASFİ KARATOSUN

  2. Tez No

    886696

    Classification of knee osteoarthritis severity using deep learning with fully supervised and semi-supervised-based approaches

    Derin öğrenme ile tamamen denetimli ve yarı-denetimli yaklaşım tabanlı diz osteoartrit şiddetinin sınıflandırılması

    İLKNUR AKTEMUR

    Yüksek Lisans

    İngilizce

    İngilizce

    2024

    Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve Kontrolİstanbul Teknik Üniversitesi

    Bilgisayar Mühendisliği Ana Bilim Dalı

    DOÇ. DR. İLKAY ÖKSÜZ

  3. Tez No

    559423

    İnsan hareketi izleme teknolojilerinin karşılaştırmalı değerlendirmesi

    Comparative assessment of human motion monitoring technologies

    CEMAL FATİH KUYUCU

    Yüksek Lisans

    Türkçe

    Türkçe

    2019

    Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve Kontrolİstanbul Teknik Üniversitesi

    Bilgisayar Mühendisliği Ana Bilim Dalı

    DR. ÖĞR. ÜYESİ GÖKHAN İNCE

  4. Tez No

    288778

    15-19 yaş bayan voleybol oyuncularında 8 haftalık antrenman programı boyunca uygulanan Kinesio® tape bantlamanın kuvvet ve esneklik performansına etkileri

    The effects of 8 weeks of Kinesio taping during the training period in muscle strength and flexibilty in 15-19 year-old women volleyball players

    İLKAY KOÇ

    Yüksek Lisans

    Türkçe

    Türkçe

    2011

    SporGazi Üniversitesi

    Beden Eğitimi ve Spor Ana Bilim Dalı

    DOÇ. DR. NEVİN ATALAY GÜZEL

  5. Tez No

    157202

    Patellofemoral ağrı sendromlu hastalarda diz ekleminin propriosepsiyon duyusunun değerlendirilmesi

    Knee joint proprioception assessment in patients with patellofemoral pain syndrome

    BURCU KANER ADALMAZ

    Tıpta Uzmanlık

    Türkçe

    Türkçe

    2004

    Fiziksel Tıp ve RehabilitasyonDokuz Eylül Üniversitesi

    Fiziksel Tıp ve Rehabilitasyon Ana Bilim Dalı

    Y.DOÇ.DR. ÇİĞDEM BİRCAN

Geri Dön