虛擬實境網路機器人遠端操控系統：視覺與攝影機校正

Abstract

在一些具有危險性或是未知的環境中，我們常常使用機器人以遠端操控的方式來完成任務。由於網路與微處理器的高度發展，使得遙距操控系統的效能大幅提高，而如果能使操作者更真切地感覺到遠端環境，必定可以提高操控性及實現度。以我們先前系統而言，在將虛擬實境技術引入遠端操控系統中時，當環境發生變化時，虛擬實境必需手動調整以維護其一致性，因此在本論文中，我們發展出一視覺與定位系統，藉由受控端的視覺系統來找出物品在影像中的位置，並且結合定位系統找出物品與機器人間的實際空間關係，再透過網路傳輸資訊，更新主控端的虛擬實境系統，使其能自動同步地跟隨實際環境變化。

In dangerous or unknown environments, we often teleoperate robot manipulators to execute the task. Due to the functions rapid development of networks and microprocessors, the functions of the teleoperation systems become much stronger. When the users can realistically feel the remote environment, better control performance may be achieved. In our previous system, humans must adjust the VR-system each time the environments vary. To let the adjustment be performed automatically in this thesis, we develop a vision and calibration system: At the remote side, the vision system is used to find out the positions of the objects in the images and the calibration system to find out the spacial relations between the objects and the robots. At the operator side, the VR-system is updated using the information of the objects through networks. Thus, we make the VR-system follow the variation of the real environment automatically and synchronously.