Towards Zero-Shot Point Cloud Registration Across Diverse Scales, Scenes, and Sensor Setups

2026-01-06

arXiv.org

AI Summary

This work addresses the limited generalization of existing point cloud registration methods in zero-shot scenarios, which often rely on dataset-specific hyperparameter tuning or retraining. We propose BUFFER-X, the first zero-shot registration framework that requires neither training nor manual parameter adjustment. It automatically estimates hyperparameters via geometric bootstrapping, replaces learned keypoint detectors with distribution-aware farthest point sampling, and ensures scale consistency through local coordinate normalization. Coupled with multi-scale matching and an efficient pose solver, BUFFER-X achieves robust cross-domain registration. Evaluated on a comprehensive benchmark encompassing 12 datasets—including object, indoor, outdoor, and heterogeneous LiDAR scans—it demonstrates state-of-the-art performance. A lightweight variant, BUFFER-X-Lite, further reduces computational time by 43% without sacrificing accuracy.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

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

Abstract

Some deep learning-based point cloud registration methods struggle with zero-shot generalization, often requiring dataset-specific hyperparameter tuning or retraining for new environments. We identify three critical limitations: (a) fixed user-defined parameters (e.g., voxel size, search radius) that fail to generalize across varying scales, (b) learned keypoint detectors exhibit poor cross-domain transferability, and (c) absolute coordinates amplify scale mismatches between datasets. To address these three issues, we present BUFFER-X, a training-free registration framework that achieves zero-shot generalization through: (a) geometric bootstrapping for automatic hyperparameter estimation, (b) distribution-aware farthest point sampling to replace learned detectors, and (c) patch-level coordinate normalization to ensure scale consistency. Our approach employs hierarchical multi-scale matching to extract correspondences across local, middle, and global receptive fields, enabling robust registration in diverse environments. For efficiency-critical applications, we introduce BUFFER-X-Lite, which reduces total computation time by 43% (relative to BUFFER-X) through early exit strategies and fast pose solvers while preserving accuracy. We evaluate on a comprehensive benchmark comprising 12 datasets spanning object-scale, indoor, and outdoor scenes, including cross-sensor registration between heterogeneous LiDAR configurations. Results demonstrate that our approach generalizes effectively without manual tuning or prior knowledge of test domains. Code: https://github.com/MIT-SPARK/BUFFER-X.

Problem

• zero-shot generalization
• point cloud registration
• cross-domain transferability
• scale inconsistency
• sensor heterogeneity

Innovation

• zero-shot generalization
• training-free registration
• geometric bootstrapping
• distribution-aware sampling
• multi-scale matching