Task-oriented Grasping in Object Stacking Scenes with CRF-based Semantic Model | IEEE Conference Publication | IEEE Xplore