BUFFER-X: Towards Zero-Shot Point Cloud Registration in Diverse Scenes

2025-03-11

AI Summary

This paper addresses the poor generalization of deep learning-based point cloud registration methods by proposing a zero-shot cross-domain registration framework to overcome three key bottlenecks: (1) environment-dependent voxel size and search radius; (2) insufficient out-of-distribution robustness of learned keypoint detectors; and (3) scale inconsistency arising from raw coordinates. Methodologically, we introduce the first adaptive voxel/search-radius estimation strategy, employ farthest-point sampling (FPS) for learning-free keypoint extraction, and propose block-wise scale normalization coupled with hierarchical multi-scale inlier filtering. We construct a comprehensive benchmark spanning 11 heterogeneous indoor and outdoor datasets. Our framework achieves significant performance gains without requiring target-domain priors, retraining, or manual hyperparameter tuning. The source code is publicly available.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Search and Optimization: Learning to Search

Application Category

• Security and Privacy: Large-scale security measurements
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments

Abstract

Recent advances in deep learning-based point cloud registration have improved generalization, yet most methods still require retraining or manual parameter tuning for each new environment. In this paper, we identify three key factors limiting generalization: (a) reliance on environment-specific voxel size and search radius, (b) poor out-of-domain robustness of learning-based keypoint detectors, and (c) raw coordinate usage, which exacerbates scale discrepancies. To address these issues, we present a zero-shot registration pipeline called BUFFER-X by (a) adaptively determining voxel size/search radii, (b) using farthest point sampling to bypass learned detectors, and (c) leveraging patch-wise scale normalization for consistent coordinate bounds. In particular, we present a multi-scale patch-based descriptor generation and a hierarchical inlier search across scales to improve robustness in diverse scenes. We also propose a novel generalizability benchmark using 11 datasets that cover various indoor/outdoor scenarios and sensor modalities, demonstrating that BUFFER-X achieves substantial generalization without prior information or manual parameter tuning for the test datasets. Our code is available at https://github.com/MIT-SPARK/BUFFER-X.

Problem

• Limitations in generalization of point cloud registration methods.
• Dependence on environment-specific parameters and poor out-of-domain robustness.
• Challenges with scale discrepancies and manual parameter tuning.

Innovation

• Adaptive voxel size and search radius determination
• Farthest point sampling replaces learned keypoint detectors
• Patch-wise scale normalization for consistent coordinate bounds