Is Image-based Object Pose Estimation Ready to Support Grasping?

TL;DR

This paper evaluates the accuracy of single RGB image-based 6-DoF object pose estimators and investigates their effectiveness for robotic grasping through simulated experiments, providing new insights into their practical usability.

Contribution

It introduces a framework for assessing pose estimators' suitability for grasping and compares five open-source methods in simulation, highlighting their real-world applicability.

Findings

Pose estimators vary in accuracy and grasp success rates.

Some estimators enable effective grasping in simulation.

Insights into the limitations of current image-based pose estimation for robotics.

Abstract

We present a framework for evaluating 6-DoF instance-level object pose estimators, focusing on those that require a single RGB (not RGB-D) image as input. Besides gaining intuition about how accurate these estimators are, we are interested in the degree to which they can serve as the sole perception mechanism for robotic grasping. To assess this, we perform grasping trials in a physics-based simulator, using image-based pose estimates to guide a parallel gripper and an underactuated robotic hand in picking up 3D models of objects. Our experiments on a subset of the BOP (Benchmark for 6D Object Pose Estimation) dataset compare five open-source object pose estimators and provide insights that were missing from the literature.