Low-bandwidth (Sparse) Visual Simultaneous Localization and Mapping (Sparse-VSLAM)

Research Overview

  • We study the visual SLAM problem under sparse sampling or low-bandwidth setting. Specifically, the sampling rate of the image frames is low (due to e.g., low-bandwith constraint) and the overlap between adjacent frames is small.
  • The main technical challenges include:

    (1) locally, difficulty in obtaining reliable inter-image correlation and matching under small partial overlap,

    (2) globally, large search space to compose global solution from unreliable local alignments, and

    (3) the efficiency requirement for online SLAM computation.

Publications

A Multi-frame Graph Matching Algorithm for Low-bandwidth RGB-D SLAM.

S. Zheng, J. Hong, K. Zhang, B. Li, and X. Li

Computer-aided Design (CAD), vol. 78, pp. 107-117, 2016.

[Paper][Bib]



Sparse3D: A New Global Model for Matching Sparse RGB-D Dataset with Small Inter-frame Overlap.

C. Le, and X. Li

Computer-aided Design (CAD), 102:33-43, 2018.

[Paper][Bib] [Codes] [Demo]



A Hardware-adaptive Deep Feature Matching Pipeline for Real-time 3D reconstruction.

S. Zheng, Y. Wang, B. Li, and X. Li

Computer-aided Design (CAD), Vol. 132, Article 102984, 2021.

[Paper] [Bib]



Locations of visitors to this page
Web Page Hit Counter