Caging is a method to make an object inescapable from a closed region by rigid bodies such as robot hands. In this paper, we test achievement of 3D multifingered caging by exploring any regions completely closed by obstructions composed of robot bodies in six dimensional configuration space. Instead of mathematical description of the obstructions and the closed regions, voxelized configuration space and labeled connections of them are adopted for approximated exploration, to avoid huge computational costs. If there are any closed regions and the target object is present in one of them, caging constraint can seem to be achieved. As a result, our proposed exploration method successfully worked to examine achievement of caging in both 2D and 3D scenes. As for 3D caging, we proposed a method to test in six dimensional configuration space using uniformly-sampled quaternion for description of orientation of object. Finally, we showed a table of relationship between the number of posture and processing time.