Abstract:Pose estimation is a crucial technique to achieve robot localization using natural features for an RGB-D vision navigation system. However, it still suffers from the impoverishment of robustness and accuracy in complex workspaces. A robust real-time pose estimation approach using hybrid point and twin line reprojection is proposed to allow accurate localization for a vision- based robot under various challenging scenarios, including low texture, low structure, occlusion, and viewpoint and illumination changes. First, an improved adaptive threshold (IAT)-based oriented fast and rotated brief (ORB) feature detection method is developed to extract sufficient IAT-ORB point features that are matched by a multistage K-nearest neighbor (KNN)-ratio-test criterion (RTC)-cosine similarity (CS)-random sample consensus (RANSAC) (KRCR) refinement feature matching strategy to generate reliable point matches. Then, a robust EDLines line feature matching method is developed by integrating dynamic matching (DM) and median absolute deviation (MAD) to obtain fine line matches. Furthermore, the hybrid point reprojection errors are built from the fusion of 3-D-2-D and 3-D-3-D point pairs, and the twin line reprojection errors are constructed by integrating virtual right-eye lines generated by depth measurement, which both are incorporated into a novel unified errors' optimization model to achieve optimal motion estimation by minimizing the reprojection errors of points and lines concurrently. Finally, the extensive experiments verify the robustness, accuracy, and real-time performance of the proposed approach on the public datasets and our real indoor scenarios. The quantitative results demonstrate that our proposed approach gains more accurate and stable localization than other state-ofthe-art methods under challenging environments.

A 3DoF Pose Estimation Method for Multi-Trolley from a Single RGB Image.

3D Point-to-Keypoint Voting Network for 6D Pose Estimation

ContourPose: Monocular 6D Pose Estimation Method for Reflective Texture-Less Metal Parts

6D pose estimation of 3D objects in scenes with mutual similarities and occlusions

Zero-Shot 3d Pose Estimation of Unseen Object by Two-Step Rgb-D Fusion

A 3D Keypoints Voting Network for 6DoF Pose Estimation in Indoor Scene

PA-Pose: Partial Point Cloud Fusion Based on Reliable Alignment for 6D Pose Tracking

In-Rack Test Tube Pose Estimation Using RGB-D Data

Robust Pose Estimation via Hybrid Point and Twin Line Reprojection for RGB-D Vision Navigation

Leaping from 2D Detection to Efficient 6DoF Object Pose Estimation.

A Two-Stage Monocular Vision Detection Method for 6D Pose Estimation in Multi-Heterogeneous Robot Systems

3D Robot Pose Estimation from 2D Images

A grasping posture estimation method based on 3D detection network

Research on Model-Free 6D Object Pose Estimation Based on Vision 3D Matching.

Semantic keypoint-based pose estimation from single RGB frames

Single-Camera Multi-View 6DoF pose estimation for robotic grasping

End-to-End 6dof Pose Estimation from Monocular RGB Images

RDPN6D: Residual-based Dense Point-wise Network for 6Dof Object Pose Estimation Based on RGB-D Images

SO-Pose: Exploiting Self-Occlusion for Direct 6D Pose Estimation

6DoF Pose Estimation of Transparent Object from a Single RGB-D Image

Robust Estimation of 3D Human Poses from a Single Image