本篇論文主旨在於實現機械手掌結合電腦視覺作抓取控制。配合安裝在機器人上的雙眼攝影機擷取影像,透過追蹤演算法(CAMSHIFT)做動態視覺分析使機器人能夠在三維空間中尋找目標物,並且利用逆向運動學理論操控機械手臂的抬舉使機器手掌移動至適當的位置接著控制機械手掌完成抓取動作。論文中使用 CAMSHIFT 追蹤演算法結合座標校正法,以提升目標物位置推算的準確率,經過實際的操作驗證了這個方法的可行性。成果可以做到人類生活上常需要的動作,增加了機器人實用的服務機能。
In this thesis, a five-finger robot hand mounted on a four-degree manipulator of an intelligent mobile robot is implemented for object grabbing based on tow-camera vision. The robot analyzes the dynamic vision acquired from stereo cameras and tracks the object in the three-dimension space utilizing CAMSHIFT algorithm. In addition, the inverse and forward kinematics is implemented to control the movement of the manipulator to appropriate position and object-grabbing of the robot hand. Moreover, a coordinate estimation correction method is developed to improve the accuracy of position calculation. The proposed methodology is proved feasible in several experiments.