Search papers, labs, and topics across Lattice.
This paper introduces PartialBiGrasp, a dual-arm grasp generation framework that operates on partial point cloud observations to infer local geometric properties essential for grasping complex objects. By employing convolutional occupancy networks, the model learns to identify graspable regions and generate force-closure compliant grasp pairs, even in the presence of incomplete geometric information. Evaluations show that PartialBiGrasp achieves robust and physically stable grasp generation in both simulated and real-world scenarios, addressing a significant limitation of prior methods that rely on complete point clouds.
PartialBiGrasp enables dual-arm robots to grasp complex objects with only partial geometric information, achieving stability where previous methods fail.
Dual-arm robotic grasping is essential for manipulating large, heavy, and geometrically complex objects that cannot be reliably handled using a single manipulator. These large objects often contain only sparse graspable regions determined by local geometric properties such as thickness, edge structure, and gripper clearance. Prior bimanual grasping methods assume access to a full point cloud of the object which inherently contains this geometric information, but may not be accessible in real scenarios. This work proposes PartialBiGrasp, a dual-arm grasp generation framework that operates directly on partial point cloud observations. Our model learns geometric features implicitly through convolutional occupancy networks, enabling local reasoning about graspability, collision-free contact regions, and object thickness. We leverage this understanding to generate force-closure compliant grasp pairs, which are further refined using a sampling-based optimization to correct for ambiguity caused by incomplete geometry. We evaluate our approach using analytical force-closure metrics, large-scale simulation experiments, and real-world robot evaluations on noisy partial point clouds of novel objects, demonstrating robust and physically stable dual-arm grasp generation.
