Abstract

In this paper, we present Sim-Grasp, a robust 6-DOF two-finger graspingsystem that integrates advanced language models for enhanced objectmanipulation in cluttered environments. We introduce the Sim-Grasp-Dataset,which includes 1,550 objects across 500 scenarios with 7.9 million annotatedlabels, and develop Sim-GraspNet to generate grasp poses from point clouds. TheSim-Grasp-Polices achieve grasping success rates of 97.14% for single objectsand 87.43% and 83.33% for mixed clutter scenarios of Levels 1-2 and Levels 3-4objects, respectively. By incorporating language models for targetidentification through text and box prompts, Sim-Grasp enables bothobject-agnostic and target picking, pushing the boundaries of intelligentrobotic systems.

 

Quick Read (beta)