Learning Object Relations with Graph Neural Networks for Target-Driven Grasping in Dense Clutter

TL;DR

This paper introduces a graph neural network-based system for target-driven grasping in dense clutter, leveraging object relations and shape completion to improve grasp success rates in real-world robotic scenarios.

Contribution

The paper presents a novel G2N2 model that evaluates object relations in a grasp graph and a shape completion method to enhance grasp sampling, advancing dense clutter grasping capabilities.

Findings

Achieves 77.78% grasping accuracy in real-world dense clutter scenarios.

Outperforms baseline methods by over 15% in grasp success rate.

Demonstrates effectiveness of relation-aware grasp evaluation and shape completion in robotics.

Abstract

Robots in the real world frequently come across identical objects in dense clutter. When evaluating grasp poses in these scenarios, a target-driven grasping system requires knowledge of spatial relations between scene objects (e.g., proximity, adjacency, and occlusions). To efficiently complete this task, we propose a target-driven grasping system that simultaneously considers object relations and predicts 6-DoF grasp poses. A densely cluttered scene is first formulated as a grasp graph with nodes representing object geometries in the grasp coordinate frame and edges indicating spatial relations between the objects. We design a Grasp Graph Neural Network (G2N2) that evaluates the grasp graph and finds the most feasible 6-DoF grasp pose for a target object. Additionally, we develop a shape completion-assisted grasp pose sampling method that improves sample quality and consequently grasping efficiency. We compare our method against several baselines in both simulated and real settings. In real-world experiments with novel objects, our approach achieves a 77.78% grasping accuracy in densely cluttered scenarios, surpassing the best-performing baseline by more than 15%. Supplementary material is available at https://sites.google.com/umn.edu/graph-grasping.