arXiv subjects

Congjia Chen

Publications and source records attributed to Congjia Chen.

4 recordsLinked to original sources

arXiv2026

Beyond Relative Geometry: Metric-Aware Geometry Perception for Robotics

Recent embodied models increasingly leverage geometric representations to improve spatial reasoning and robotic manipulation. However, existing reconstruction methods only reconstruct relative geometry with arbitrary scales, causing predicted object dimensions and spatial distances to vary across scenes, viewpoints, and input configurations. This inconsistency prevents geometric perception from being directly aligned with robotic actions defined on the real-world scale. To address this limitation, we propose Metric-Aware Geometry Perception (MAGP), an end-to-end, plug-and-play framework for metric geometry reconstruction that can be seamlessly integrated into robotic policies. At its core, Metric Scale Equivariant Augmentation encourages the model to reconstruct metric geometry from camera parameters and depth observations, ensuring that the reconstructed geometry follows the metric scale specified by observations. Flexible Metric Conditioning further enables MAGP to support arbitrary view counts and combinations of camera and depth inputs, improving robustness to heterogeneous robotic sensing configurations. Together, these designs produce geometrically consistent reconstructions with stable object dimensions and spatial distances across scenes and sensing conditions. Experiments on ETH3D, MegaDepth, and ScanNet++ demonstrate that MAGP maintains strong relative geometry accuracy while reducing the absolute error by over an order of magnitude, from 2.01m to 0.07m. When integrated into multiple robotic policies, MAGP consistently improves performance on LIBERO, RoboTwin, and zero-shot LIBERO-Plus, with gains of up to 6.26% on RoboTwin. These results demonstrate the effectiveness and generalizability of metric geometry for robotic manipulation.

cs.RO↗

arXiv2025

ViGG: Robust RGB-D Point Cloud Registration using Visual-Geometric Mutual Guidance

Point cloud registration is a fundamental task in 3D vision. Most existing methods only use geometric information for registration. Recently proposed RGB-D registration methods primarily focus on feature fusion or improving feature learning, which limits their ability to exploit image information and hinders their practical applicability. In this paper, we propose ViGG, a robust RGB-D registration method using mutual guidance. First, we solve clique alignment in a visual-geometric combination form, employing a geometric guidance design to suppress ambiguous cliques. Second, to mitigate accuracy degradation caused by noise in visual matches, we propose a visual-guided geometric matching method that utilizes visual priors to determine the search space, enabling the extraction of high-quality, noise-insensitive correspondences. This mutual guidance strategy brings our method superior robustness, making it applicable for various RGB-D registration tasks. The experiments on 3DMatch, ScanNet and KITTI datasets show that our method outperforms recent state-of-the-art methods in both learning-free and learning-based settings. Code is available at https://github.com/ccjccjccj/ViGG.

cs.CV↗

arXiv2025

DINOReg: Strong Point Cloud Registration with Vision Foundation Model

Point cloud registration is a fundamental task in 3D computer vision. Most existing methods rely solely on geometric information for feature extraction and matching. Recently, several studies have incorporated color information from RGB-D data into feature extraction. Although these methods achieve remarkable improvements, they have not fully exploited the abundant texture and semantic information in images, and the feature fusion is performed in an image-lossy manner, which limit their performance. In this paper, we propose DINOReg, a registration network that sufficiently utilizes both visual and geometric information to solve the point cloud registration problem. Inspired by advances in vision foundation models, we employ DINOv2 to extract informative visual features from images, and fuse visual and geometric features at the patch level. This design effectively combines the rich texture and global semantic information extracted by DINOv2 with the detailed geometric structure information captured by the geometric backbone. Additionally, a mixed positional embedding is proposed to encode positional information from both image space and point cloud space, which enhances the model's ability to perceive spatial relationships between patches. Extensive experiments on the RGBD-3DMatch and RGBD-3DLoMatch datasets demonstrate that our method achieves significant improvements over state-of-the-art geometry-only and multi-modal registration methods, with a 14.2% increase in patch inlier ratio and a 15.7% increase in registration recall. The code is publicly available at https://github.com/ccjccjccj/DINOReg.

cs.CV↗

arXiv2024

RGBD-Glue: General Feature Combination for Robust RGB-D Point Cloud Registration

Point cloud registration is a fundamental task for estimating rigid transformations between point clouds. Previous studies have used geometric information for extracting features, matching and estimating transformation. Recently, owing to the advancement of RGB-D sensors, researchers have attempted to combine visual and geometric information to improve registration performance. However, these studies focused on extracting distinctive features by deep feature fusion, which cannot effectively solve the negative effects of each feature's weakness, and cannot sufficiently leverage the valid information. In this paper, we propose a new feature combination framework, which applies a looser but more effective combination. An explicit filter based on transformation consistency is designed for the combination framework, which can overcome each feature's weakness. And an adaptive threshold determined by the error distribution is proposed to extract more valid information from the two types of features. Owing to the distinctive design, our proposed framework can estimate more accurate correspondences and is applicable to both hand-crafted and learning-based feature descriptors. Experiments on ScanNet and 3DMatch show that our method achieves a state-of-the-art performance.

cs.CV↗