RGB-Dハンドアイカメラを有する2指ハンドによる任意形状のワークのピッキング

―把持位置検出の計算効率化およびワークとハンドの干渉回避による把持成功率の改善―

抄録

We improve a method for grasping arbitrary shape objects applicable to multijoint manipulators with a two-finger hand and an RGB-D hand-eye camera. The computational cost for detecting grasping positions is reduced by limiting computing area. A collision avoidance algorithm between the hand and objects is also newly implemented. Our new implementation is evaluated with experiments in which an actual manipulator picks up each object from their pile. Experimental results suggest that not only the robot come to grasp the object within 11 seconds, but also it is able to grasp from a pile of work.