Title

Learning Dissipative Dynamics : A Neural Network Approach Integrated with Lagrangian Mechanics

Alternative Title

소산 동역학 학습: 라그랑주 역학과 통합된 신경망 접근법

DGIST Authors

Hoyeong Yeo ; Sehoon Oh ; Sukho Park

Advisor

오세훈

Co-Advisor(s)

Sukho Park

Issued Date

2025

Awarded Date

2025-08-01

Type

Thesis

Description

Legged Robot, Lagrangian Mechanics, Dynamic Friction, Neural Network, Force Observer

Table Of Contents

Ⅰ. INTRODUCTION 1

ⅠI. MIMO HYSTERETIC FRICTION-AWARE LAGRANGIAN-BASED NETWORK 5
2.1 Lagrangian-based Neural Networks 8
2.2 Dissipative Energy of Non-conservative System 12
2.3 Proposed Method : Mysteric Net (M-Net) 18

ⅠII. EXPERIMENTAL VALIDATION OF SYSTEM MODELING 20
3.1 System Information : Legged Robot 20
3.2 Experiment Design and Data Acquisition 22
3.3 Results 25

ⅠV. MYSTERIC NET-BASED FORCE OBSERVER 30
4.1 Experiment Design and Data Acquisition 31
4.2 Results 32

V. CONCLUSION 37

REFERENCE 38

SUMMARY (in Korean) 40

URI

https://scholar.dgist.ac.kr/handle/20.500.11750/59823
http://dgist.dcollection.net/common/orgView/200000889953

DOI

10.22677/THESIS.200000889953

Degree

Master

Department

Department of Robotics and Mechatronics Engineering

Publisher

DGIST

Show Full Item Record