機器手掌抓取薄片物體的策略

Abstract

在人類生活的環境中，物體放置的位置會隨著人類的使用習慣有所變動，而物體出現的位置並不總是在可方便拿取的位置，例如：名片放在桌面上，直接用手指攫取名片的方式，並不是一個最有效率的方式。因此，本論文旨在探討如何藉由機器手掌與機器手臂的合作，成功拿取輕薄物體。並透過影像的定位與追蹤，進行拿取物體的前置動作，調整物體在環境中的狀態，來提高拿取物體的成功率。本論文之理論架構主要分為兩個部分：第一部分是經由力量控制來完成拿取薄片的任務，此力量控制的功能在於結合組抗控制的適應性，讓機器手得以適應物體與環境，並在拖曳的過程中，使用模糊控制進行力量的補償，以改善物體可能滑落的情形。第二部分是機器手掌與手臂系統的規劃，針對拿取薄片的任務，建立三個階段的任務規劃，分別是：適應物體、拖曳物體與拿取物體。最後，在已配置機器手掌NTU-Hand V和NTU的六軸機器手臂上，完成拿取輕薄物體等任務，以驗證整個系統的控制與規劃。相信此手掌與手臂系統之發展，可運用到日常生活中，進行抓取及操作物體等任務，展現機器手掌的靈巧性與多工性。

In the environment of human life, object placement may be changed in terms of human use, and the location of an object might not always appear in the most convenient place for reaching a desired grasp. For example, directly reaching and grasping a card located on a desk is not the most efficient way to grasp it. This thesis is aimed to use a robotic hand and arm to grasp a thin object successfully. Through image navigation and tracking, information about the object can be obtained to perform a pre-grasp manipulation to adjust the placement of the object and improve the success rate of the grasp. The thesis is divided into two parts. The first part discusses force control, which takes advantage of impedance control to make the fingers comply with the object and the environment, and fuzzy logic control to manage the force for handling slippage during the sliding manipulation. The second part is focused on the planning of the hand-arm system. It can be established in three stages: preshape adaptation, pre-grasp manipulation, and final grasp. Finally, we perform the task of grasping a thin object with an NTU Hand V and six-DOF NTU Arm to demonstrate the entire planning and control process.