English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

6-DoF object-agnostic grasping in unstructured environments is a critical yetchallenging task in robotics. Most current works use non-optimized approachesto sample grasp locations and learn spatial features without concerning thegrasping task. This paper proposes GraNet, a graph-based grasp pose generationframework that translates a point cloud scene into multi-level graphs andpropagates features through graph neural networks. By building graphs at thescene level, object level, and grasp point level, GraNet enhances featureembedding at multiple scales while progressively converging to the idealgrasping locations by learning. Our pipeline can thus characterize the spatialdistribution of grasps in cluttered scenes, leading to a higher rate ofeffective grasping. Furthermore, we enhance the representation ability ofscalable graph networks by a structure-aware attention mechanism to exploitlocal relations in graphs. Our method achieves state-of-the-art performance onthe large-scale GraspNet-1Billion benchmark, especially in grasping unseenobjects (+11.62 AP). The real robot experiment shows a high success rate ingrasping scattered objects, verifying the effectiveness of the proposedapproach in unstructured environments.

GraNet: A Multi-Level Graph Network for 6-DoF Grasp Pose Generation in Cluttered Scenes