虛擬導引於機器人遠端操作之應用

Abstract

由於現今網路技術的發達，人類已經可以藉由操作機器人在各種嚴酷的環境下進行未知領域的探索，但是遠端操控系統也存在許多缺點，像是由於網路的延遲使得操作者無法即時得到遠方機器人的資訊，另外攝影機回傳的影像限制，造成操作者難以判斷物體的遠近關係，這些問題使得遠端操控在實際應用上產生困難，因而有學者提出利用虛擬場景(Virtual environment)配合虛擬導引(Virtual guidance)來輔助操作者進行遠端操控，虛擬導引乃透過虛擬場景中提供的視覺、力覺等對操作者提供輔助，在操作上的穩定性以及效率都有相當大的幫助。對於虛擬導引的建構方法，已有相當多的研究及著作，本文提出一鎖定形式虛擬導引方法，該方法利用攝影機提供的平面影像，配合力輔助的導引，使操作者在操作過程可矯正夾爪位置，彌補定位上的不精確，進而完成工作。利用所提出方法，我們建構了一網路連結遠端操控機器人系統進行實地操作，並獲得不錯的實驗結果，證明鎖定式虛擬導引在提昇效率以及穩定性方面，都可以發揮很好的效果。

Rapid development in computer network highly enhances the capability of teleoperation systems, people nowadays can explore an unknown area by using a remote-controlled robot. However, there are still some drawbacks in a telerobtic system.For instance, the operator can not obtain remote information fast enough due to the network time delay. Meanwhile, it is also difficult to distinguish the relative distance between objects from the planar image. Some researchers proposed using virtual environment and virtual guidance to tackles the problems. Virtual guidance acts as a guidance that aids the user in manipulating the robot via providing haptic and visual clues. In this thesis, we propose a novel virtual guidance which is more effective than those proposed previously in object grasping. The proposed virtual guidance integrates the planar image and force reflection to provide the guidance. With it, the operator can easily control the robot gripper for object grasping in the presence of imprecision in positioning. In implementation, we developed a networked VR-based telerobotic system for the proposed virtual guidance. Experiments are performed to demonstrate the effectiveness of the proposed approach.