Title

Integration of Physics-Informed Neural Networks in MPC for Optimal Control of Mobile Robot: A Comparative Performance Analysis

Alternative Title

이동 로봇의 최적 제어를 위한 물리 정보를 포함한 신경망의 모델 예측 제어 통합: 비교 성능 분석

DGIST Authors

Doyoung Kim ; Yongseob Lim ; Soon Kwon

Advisor

임용섭

Co-Advisor(s)

Soon Kwon

Issued Date

2024

Awarded Date

2024-08-01

Citation

Doyoung Kim. (2024). Integration of Physics-Informed Neural Networks in MPC for Optimal Control of Mobile Robot: A Comparative Performance Analysis. doi: 10.22677/THESIS.200000798163

Type

Thesis

Description

Physical Information Neural Networks, Model Predictive Control, Mobile Robots, Optimal Control, Self-Driving Mobile Robots

Table Of Contents

Ⅰ. Introduction 1
Ⅱ. Related Works 3
Ⅲ. Mobile Robot Kinematics and State-space Model 4
3.1 Mobile Robot Kinematics 4
3.2 State-space Model for MPC and PINN 6
Ⅳ. Problem Statement
4.1 Model Predictive Control 8
4.2 Physics-Informed Neural Network 10
4.3 PINN Training 12
4.4 Integration of PINN into the MPC Framework 14
Ⅴ. Experiment Result
5.1 PINN model Accuracy 15
Ⅵ. Conclusions, Limitations and Future Works 16

References 17
국문 초록 19

URI

http://hdl.handle.net/20.500.11750/57614
http://dgist.dcollection.net/common/orgView/200000798163

DOI

10.22677/THESIS.200000798163

Degree

Master

Department

Department of Robotics and Mechatronics Engineering

Publisher

DGIST

Show Full Item Record