虛擬實境網路機器人遠端操控系統：物體形變與機器人智慧

Abstract

由於機電技術的進步，使得結合虛擬實境的遠端操控系統其應用愈來愈廣泛，像是外太空、深海、家用機器人等。因此，機器人所要面對的環境和任務也相對的趨於複雜。就整個遠端操控系統而言，由於操作者身處整個控制迴圈中，如何讓操作者能感受到實際環境的變化，減輕其工作的負擔，是相當重要，也是我們在此論文中所要探討的。本文中，我們在虛擬實境網路機器人遠端操控系統的架構中，增加了物件的形變和機器人的智慧等效能，我們也設計更一般化的 Peg-in-Hole 實驗，利用機器人來辨別具不同軟硬度的物件，依此決定夾爪該施多少力夾起物件，再進行組裝。同時，於VR的環境中，我們也可以即時地觀察到物件表面的變化；實驗結果印証了我們所發展的新功能的成效，也讓此虛擬實境網路機器人遠端操控系統得以應用到更廣的領域，如遙控手術等。

Due to the advancement of the mechatronic technologies, the application of the VR-based teleoperation system is more popular, such as outer space, deep water, and home-robot, etc. Therefore, the robot needs to face more complex tasks and environments. As the operator is the main part in the control loop of a teleoperation system, it is important to let the operator realize the happenings in the environment and also share her (his) load. In this thesis, we enhance the networked VR-based telerobotic system previously developed in our laboratory with new functions of object deformation and robot intelligence. We have also designed a more general peg-in-hole experiment for verification. The robot will find out the stiffness of the object first, and the gripper can then determine proper force to grab the object. Meanwhile, we can also observe the object deformation process in the virtual environment. The experiment results demonstrate that the effect of the functions newly developed. Consequently, this VR-based teleoperation system can be applied broader areas, such as tele-surgery, etc.