Title: | 結合虛擬實境與機器人之遠端操控系統:力感之呈現與操控 VR/Robot Integration: Force Reflection and Manipulation |
Authors: | 周立人 Chou, Li-Ren 楊谷洋 Young Kuu-Young 電控工程研究所 |
Keywords: | 遠端操控;虛擬實境;力感回饋;機器人;最小平方法;傳輸延遲;tele-operation;VR;force reflection;telerobot;least squre method;time delay |
Issue Date: | 1997 |
Abstract: | 在進行遠端操作時,操作者必須要能明瞭遠端環境的情況才能達成理 想的操控。近年來由於虛擬實境的快速成長,利用虛擬實境的技術,我們 將可以提供給操作者更佳的臨場感,據此構想,我們發展了一套結合虛擬 實境與機器人的遠端操控系統。視覺化的虛擬環境讓操作者擁有了逼真的 感受。然而,當操作遠端機器人與環境接觸時,力的感覺是不可或缺的, 因此在本篇論文中,除了已發展的虛擬實境與機器人的遠端操控系統外, 我們的主要目的便是要在此一系統上加入力感。由於資料傳輸上的延遲與 所使用之機器人本身控制器的限制,我們發現將偵測到的力訊號直接回饋 反應給操作者,很容易造成系統的不穩定,因而我們提出了『虛擬力感回 饋』的方法,藉由即時估測環境的物理參數,根據物理模型模擬出接觸的 力量,可以使得操作者穩定而有效的操縱遠端機器人去與環境接觸並完成 順應性的工作。因為虛擬力感是由電腦產生並不需透過遠端傳輸,所以一 切情況都先在電腦的虛擬環境下預先發生了,傳輸延遲也不會再影響系統 。另外,由於環境的物理參數是在接觸環境時才即時去估測的,所以系統 可以在未知的環境下工作。實驗的結果闡明了此一系統的可行性。 In tele-operation, it is very important to reflect the situations in the remote real world to the operator faithfully. Using virtual reality (VR) technology, we propose VR-based telerobotic system, which can provide the operator more realistic feelings for tele-operation. When we operate the telerobot to interact with environments, force feeling becomes necessary for tele-operation. To complete the system, in this thesis, our main purpose is to equip force feeling on the proposed VR-based telerobotic system. Because of the transmission time delay and the limitation of the employed position-driven robot manipulator, the direct feedback force reflection may easily invoke instability. Therefore, we propose "VR force reflection" to implement force reflection in tele- operation. By estimating object stiffnesses in real time, the interactive force can be emulated and reflected to the operator. With VR force reflection, the operator can feel stable and realistic force feeling and execute compliance tasks stably and efficiently. Because the reflection force is acquired "locally", the system has the capability to ``predict" the happenings in the remote real world and the problem of time delay is then alleviated. And because the the object stiffnesses are estimated in real time, we can deal with unknown environments. The experimental results demonstrate the effectiveness of the proposed system. |
URI: | http://140.113.39.130/cdrfb3/record/nctu/#NT860591016 http://hdl.handle.net/11536/63192 |
Appears in Collections: | Thesis |