Search papers, labs, and topics across Lattice.
This paper reformulates non-prehensile manipulation by optimizing a graspability objective rather than targeting a predefined object pose, addressing the challenge of multiple graspable configurations. By constructing a graspability field that quantifies the suitability of object configurations for successful grasp execution, the authors enable a closed-loop manipulation-to-grasp pipeline driven by a single reinforcement learning policy. Experiments demonstrate that this approach effectively reconfigures objects into graspable states and correlates predicted graspability with real-world grasp success, enhancing the efficiency of robotic manipulation tasks.
A novel graspability field allows robots to autonomously determine optimal object configurations for grasping, eliminating the need for predefined poses.
Non-prehensile manipulation is often used as a preparatory step for robotic grasping, yet existing approaches typically require a predefined target object pose. In practice, however, objects admit multiple graspable configurations and the desired pose is not known in advance. We reformulate non-prehensile manipulation for grasping as optimizing an object centric graspability objective rather than reaching a specific pose. We construct a graspable set from synthesized grasps and define a graspability field that measures how suitable an object configuration is for successful grasp execution. The scalar measure provides a dense learning signal for reinforcement learning and determines when to terminate manipulation. This yields a closed-loop manipulation-to-grasp pipeline driven by a single policy. Experiments in simulation and on a real robot show that the policy reliably reconfigures objects into graspable states and transitions to grasping without external planners or manually specified stopping conditions. The predicted graspability distance correlates with real world grasp success, which indicates that the learned representation captures grasp feasibility of object configurations.
