Motion control of deep sea vehicle 'OTOHIME': modeling with neural network

被引：2

作者：

Fujii, S. ^{[1
]}

Tahara, J. ^{[2
]}

Zhang, F. ^{[2
]}

Koike, M. ^{[2
]}

Ohta, Y. ^{[3
]}

Watanabe, Y. ^{[3
,4
]}

机构：

[1] Tokyo Univ Marine Sci & Technol, Grad Sch Marine Sci & Technol, Tokyo, Japan

[2] Tokyo Univ Marine Sci & Technol, Dept Marine Elect & Mech Engn, Tokyo, Japan

[3] Japan Agcy Marine Earth Sci & Technol, Marine Technol & Engn Ctr, Yokosuka, Kanagawa, Japan

[4] Japan Agcy Marine Earth Sci & Technol, Inst Marine Earth Explorat & Engn, Yokosuka, Kanagawa, Japan

来源：

ADVANCED ROBOTICS | 2021年 / 35卷 / 24期

关键词：

Underwater robotics; modeling; neural network; nonlinear autoregressive external; long short-term memory; IDENTIFICATION;

D O I：

10.1080/01691864.2021.1985606

中图分类号：

TP24 [机器人技术];

学科分类号：

080202 ; 1405 ;

摘要：

Deep sea survey vehicle 'OTOHIME' is one of the underwater vehicle owned in JAMSTEC and its automatic navigation system is developing now. To construct the control system, vehicle motion should be modeled at first. In this study, we treated a dynamic equation model and two types of neural network models and compared their performance. The latter showed rather better results than the former. Then we decide to use the neural network model to design the control system as the next step.

引用

页码：1500 / 1512

页数：13

共 22 条

[1]

[Anonymous], 1995, JAPAN SOC NAVAL ARCH

[2] A robust neural network approximation-based prescribed performance output-feedback controller for autonomous underwater vehicles with actuators saturation [J].