Development of Motion Sensory Systems for Estimating the Motion States and the Control of an Underwater Operation System (III)

瀏覽統計 Email 通知 RSS Feed

簡歷

基本資料

Project title

Code/計畫編號

NSC94-2611-E019-001

Translated Name/計畫中文名

水下感測元件及系統之研發應用於水下作業---子計畫IV：運動感測系統之研發用於水下作業系統之運動狀態估測與控制(III)

Project Coordinator/計畫主持人

Chen-Chou Lin

Funding Organization/主管機關

National Science and Technology Council

Department/Unit

Department of Mechanical and Mechatronic Engineering

Website

https://www.grb.gov.tw/search/planDetail?id=1147591

Year

2005

Start date/計畫起

01-08-2005

Expected Completion/計畫迄

01-07-2006

Bugetid/研究經費

809千元

ResearchField/研究領域

土木水利工程
電子電機工程

本成果報告為整合型計畫之子計畫四第三年的研究成果。子計畫之總目標為研發運動感測系統用於量測與估算水下作業系統運動狀態之相關技術，並發展以感測器為基礎之水下作業系統運動控制器。本年的目標為發展類神經網路為基礎之力迴授定位，使水下機械臂之定位具有強健性。本研究中提出以下構想：先利用視覺系統進行粗略之位置量測，令機械臂之終端器移動至插配任務之大約位置，再搭配力量回授信號與類神經網路演算所求得之位置資訊，命令機械臂進行細部位置調整，最後完成自動化插配任務。本研究中我們主要探討運用力量回授於自動化插配任務，該插配系統乃使用五軸機械臂，並將六軸力量扭矩感測器安裝於機械臂腕部位置，由力量扭矩感測器所回授之信號透過類神經網路來計算出夾爪所須修正之位置誤差量後，再驅動機械臂完成插配任務。其主要優點為不需建立動態模型進行控制，並可直接對機械臂驅動器輸入夾爪位置誤差量，驅動機械臂完成插配任務。在水下進行夾具放置或組裝作業時，本實驗結果顯示定位誤差分佈仍屬可接受之範圍，因此我們認為採用類神經網路進行力迴授定位為可行之策略，但仍有改善之空間。 The research report contains the result of the third-year subproject of an integrated project. The main objective is to develop the motion sensory systems for measuring and estimating the motion states of the underwater operation system, and to develop the sensor-based motion controller of the underwater operation system. The topic of this year is to develop the force feedback positioning scheme based on the neural network. In this research we propose the following idea: by using the machine vision, a rough position information can be obtained in advanced, and the robot is commanded to a nearby position of the desired position. Then force information is measured and fed into the neural network system to obtain a new modified position. The positions of the robot end effector can thus be modified sequentially and finally converge to the desired position and complete the peg-in-hole task. The advantage of the above scheme is that a precise model of the robot and the environment is not required and the result can be more robust. The result shows that the position error of the peg-in-hole experiment using neural network force feedback scheme is acceptable for underwater operation such as clamp mounting. Hence the force feedback position using neural network can be regarded as a feasible scheme, but further improvement is required

Keyword(s)

水下作業
力量迴授系統
類神經網路
差配作業
定位誤差

DSpace CRIS

Development of Motion Sensory Systems for Estimating the Motion States and the Control of an Underwater Operation System (III)

基本資料

Description