PointNetLK Revisited

TL;DR

This paper revisits PointNetLK, demonstrating that incorporating an analytical Jacobian enhances its generalization ability in point cloud registration, outperforming state-of-the-art methods in mismatched conditions and maintaining competitiveness on real-world data.

Contribution

The paper shows that adding an analytical Jacobian to PointNetLK significantly improves its generalization in challenging registration scenarios.

Findings

Outperforms state-of-the-art in mismatched conditions

Maintains competitive results on real-world data

Enhances generalization through analytical Jacobian

Abstract

We address the generalization ability of recent learning-based point cloud registration methods. Despite their success, these approaches tend to have poor performance when applied to mismatched conditions that are not well-represented in the training set, such as unseen object categories, different complex scenes, or unknown depth sensors. In these circumstances, it has often been better to rely on classical non-learning methods (e.g., Iterative Closest Point), which have better generalization ability. Hybrid learning methods, that use learning for predicting point correspondences and then a deterministic step for alignment, have offered some respite, but are still limited in their generalization abilities. We revisit a recent innovation -- PointNetLK -- and show that the inclusion of an analytical Jacobian can exhibit remarkable generalization properties while reaping the inherent fidelity benefits of a learning framework. Our approach not only outperforms the state-of-the-art in mismatched conditions but also produces results competitive with current learning methods when operating on real-world test data close to the training set.