Supermarket-6DoF: A Real-World Grasping Dataset and Grasp Pose Representation Analysis

TL;DR

Supermarket-6DoF is a real-world grasping dataset with 1500 attempts on supermarket objects, providing detailed 6-DoF grasp annotations and demonstrating the effectiveness of point cloud-based grasp representations.

Contribution

The paper introduces Supermarket-6DoF, a novel real-world dataset with comprehensive 6-DoF grasp annotations and analysis of grasp pose representations for success prediction.

Findings

Point cloud-based grasp representation outperforms quaternion encoding.

The dataset enables analysis of grasp success and stability in real-world scenarios.

Explicit gripper geometry modeling improves grasp prediction accuracy.

Abstract

We present Supermarket-6DoF, a real-world dataset of 1500 grasp attempts across 20 supermarket objects with publicly available 3D models. Unlike most existing grasping datasets that rely on analytical metrics or simulation for grasp labeling, our dataset provides ground-truth outcomes from physical robot executions. Among the few real-world grasping datasets, wile more modest in size, Supermarket-6DoF uniquely features full 6-DoF grasp poses annotated with both initial grasp success and post-grasp stability under external perturbation. We demonstrate the dataset's utility by analyzing three grasp pose representations for grasp success prediction from point clouds. Our results show that representing the gripper geometry explicitly as a point cloud achieves higher prediction accuracy compared to conventional quaternion-based grasp pose encoding.