SporeAgent: Reinforced Scene-level Plausibility for Object Pose Refinement

TL;DR

SporeAgent introduces a scene-level plausibility approach for object pose refinement, leveraging physical plausibility constraints to improve accuracy in cluttered environments, outperforming existing methods.

Contribution

It extends RL-based pose refinement by incorporating scene plausibility, reducing ambiguity and enhancing accuracy in object pose estimation.

Findings

Achieves state-of-the-art results on LINEMOD dataset.

Improves pose accuracy in cluttered scenes.

Leverages physical plausibility to reduce ambiguity.

Abstract

Observational noise, inaccurate segmentation and ambiguity due to symmetry and occlusion lead to inaccurate object pose estimates. While depth- and RGB-based pose refinement approaches increase the accuracy of the resulting pose estimates, they are susceptible to ambiguity in the observation as they consider visual alignment. We propose to leverage the fact that we often observe static, rigid scenes. Thus, the objects therein need to be under physically plausible poses. We show that considering plausibility reduces ambiguity and, in consequence, allows poses to be more accurately predicted in cluttered environments. To this end, we extend a recent RL-based registration approach towards iterative refinement of object poses. Experiments on the LINEMOD and YCB-VIDEO datasets demonstrate the state-of-the-art performance of our depth-based refinement approach.