Search papers, labs, and topics across Lattice.
This paper introduces a hybrid grasp synthesis framework that integrates a learning-based Energy-Based Model (EBM) with an analytical Iterative Closest Point (ICP) method to effectively generate robust grasps from partially observed point clouds. By employing a learned energy function as a prior within a Stein Variational Gradient Descent (SVGD) framework, the method iteratively refines grasp configurations, leading to significant improvements in grasp success rates. Evaluated on a diverse set of 67 objects, the framework achieves a 60.9% success rate, substantially surpassing existing methods such as AnyGrasp and Grasp Pose Detection, showcasing its strong generalization capabilities.
Combining learning and geometric optimization, this framework achieves a 60.9% grasp success rate, outperforming traditional methods by a significant margin.
We propose a hybrid grasp synthesis framework that combines a learning-based Energy-Based Model (EBM) with an analytical Iterative Closest Point (ICP) method to generate robust grasps from partially observed point clouds. The learned energy function acts as a prior within a Stein Variational Gradient Descent (SVGD) framework, guiding iterative refinement of grasp configurations. Evaluated on 67 objects with 5,360 grasp attempts, our method achieves an average success rate of 60.9\%, outperforming AnyGrasp (31.1\%) and Grasp Pose Detection (48.4\%) and AS-ICP (56.6\%). These results highlight the strong generalization ability of our approach and demonstrate how combining data-driven learning with geometric optimization addresses the limitations of either strategy in isolation.
NVIDIA