Rearrangement with Nonprehensile Manipulation Using Deep Reinforcement Learning
This addresses the challenge of object rearrangement in robotics without precise physical modeling, offering a practical solution for tasks where explicit planning is infeasible, though it is incremental as it builds on existing deep reinforcement learning methods.
The paper tackled the problem of rearranging objects on a tabletop using nonprehensile manipulation by learning a strategy with deep reinforcement learning based on visual feedback, achieving an 85% success rate in simulation and showing adaptability to environmental changes.
Rearranging objects on a tabletop surface by means of nonprehensile manipulation is a task which requires skillful interaction with the physical world. Usually, this is achieved by precisely modeling physical properties of the objects, robot, and the environment for explicit planning. In contrast, as explicitly modeling the physical environment is not always feasible and involves various uncertainties, we learn a nonprehensile rearrangement strategy with deep reinforcement learning based on only visual feedback. For this, we model the task with rewards and train a deep Q-network. Our potential field-based heuristic exploration strategy reduces the amount of collisions which lead to suboptimal outcomes and we actively balance the training set to avoid bias towards poor examples. Our training process leads to quicker learning and better performance on the task as compared to uniform exploration and standard experience replay. We demonstrate empirical evidence from simulation that our method leads to a success rate of 85%, show that our system can cope with sudden changes of the environment, and compare our performance with human level performance.