Combining Reinforcement Learning and Rule-based Method to Manipulate Objects in Clutter | IEEE Conference Publication | IEEE Xplore