Multimodal Grasp Planner for Hybrid Grippers in Cluttered Scenes

Grasping a variety of objects is still an open problem in robotics, especially for cluttered scenarios. Multimodal grasping has been recognized as promising strategy to improve the manipulation capabilities robotic system. This work presents novel grasp planning algorithm hybrid grippers that allows multiple modalities. In particular, planner manages two-finger grasps, single or double suction and magnetic grasps. Grasps different modalities are geometrically computed based on cuboid material properties clutter. The presented framework modular can leverage any 6D pose estimation segmentation network far they satisfy required interface. Furthermore, be applied (hybrid) gripper, provided gripper clearance, finger width, diameter. approach fast low computational burden, it uses geometric computations synthesis selection. performance system assessed with experimental campaign three scenarios increasing difficulty using YCB dataset DLR hybrid-compliant gripper.