本論文提出一種混合力量與雙眼視覺之六軸機械手臂追蹤控制系統,透過眼看手(eye-to-hand) 架構整合影像處理、雙眼視覺、影像追蹤、三維重建、六軸機械手臂控制與力量控制等技術,成功建構一套追蹤控制系統。本系統運用兩組架設於環境中之雙眼攝影機分別觀察操控者手持模擬致動器與裝置於機械手臂末端的致動器,利用影像處理擷取致動器上之特徵,藉由此特徵分別將致動器之三維座標資訊重建於攝影機座標系統下。然而欲執行雙眼視覺之機械手臂追隨控制需依據預先校準之手眼關係,將重建於攝影機三維座標系統之特徵轉換至機械手臂基底座標系統下,進而結合力量控制執行機械手臂定位與追蹤任務。在此追蹤控制系統中,操控者利用手持模擬致動器引導機械手臂末端的致動器至一未知動態曲面並進行定位控制,當機械手臂於未知動態曲面定位完成時,藉由六軸力量感測器回傳之感測資訊重建接觸平面,並利用此接觸平面執行力量控制,使得機械手臂末端的致動器與接觸平面維持一固定接觸力,進而達成即時追蹤控制。本論文利用一台六軸工業型機械手臂與固定式雙眼攝影機,已成功驗證所提出的混合力量與視覺控制方法之可行性。
This thesis presents a hybrid force and vision-based tracking control system for a six-DOF robotic manipulator. The proposed robotic control system is in eye-to-hand configuration with binocular vision and force sensing. By using binocular vision systems, both the tool held by the operator, and the other tool mounted on the end-effector of the manipulator can be observed in real-time. The 3D coordinates of the features on both tools can be reconstructed accordingly. The tool mounted on the end-effector of the manipulator can thus be controlled to follow the pose determined by the operator and then reach an unknown surface. Based on the measurements form a six-axis force sensor, the tool is further controlled to maintain the desired contact force even when the surface is varying. The proposed hybrid force and vision-based control approach has been effectively validated with a six-DOF industrial robotic manipulator equipped with fixed binocular CCD cameras.