Plane-Informed Dynamic Visual SLAM in Low-Texture Indoor Construction Environments

Abstract Robots have been utilized increasingly in interior decoration such as bricklaying, wall plastering, and floor tiling. These tasks require precise localization, in which simultaneous localization and mapping (SLAM) technology plays a critical role. Visual SLAM recently has become an appealing and cost-effective solution with rich semantic information for scene understanding. However, indoor construction workplaces are extremely low-texture and dynamic environments, with limited and stable features for visual SLAM. Although structural SLAM can extract geometric features for localization, most of these methods rely on point and line features, which are sparse in indoor construction workplaces, in which there are only a few planes such as walls, ceilings and floors, without interior decoration providing line features. The challenge is complicated further by the dynamic environments with moving workers and machines. How to extract reliable features for visual SLAM in such environments is a critical problem for indoor construction robots that remains under-investigated. Based on the observation that planes are the dominant geometric features in indoor construction workplaces, this paper proposes a plane-informed SLAM methodology to address the challenge by leveraging accurate plane detection and plane-based static region detection. Experiments on real-world construction sites indicated consistent improvement of over 50% in localization accuracy and demonstrate stable performance across different dynamic sites, outperforming other visual SLAM systems. This work demonstrates that planes can serve as the primary structural information and can contribute to a visual SLAM system robust to low-texture and dynamic indoor construction environments.

Authors

Institutions

Publication Details

Journal
Journal of Computing in Civil Engineering
Published
2026-11-14
DOI
https://doi.org/10.1061/jccee5.cpeng-7952
Primary Topic
Robotics and Sensor-Based Localization
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Plane-Informed Dynamic Visual SLAM in Low-Texture Indoor Construction Environments

Yantao Yu, Wenyu Li, Xinyu Chen
Journal of Computing in Civil Engineering
Robotics and Sensor-Based Localization
article

Plane-Informed Dynamic Visual SLAM in Low-Texture Indoor Construction Environments

Yantao Yu, Wenyu Li, Xinyu Chen
article en

Abstract

Abstract Robots have been utilized increasingly in interior decoration such as bricklaying, wall plastering, and floor tiling. These tasks require precise localization, in which simultaneous localization and mapping (SLAM) technology plays a critical role. Visual SLAM recently has become an appealing and cost-effective solution with rich semantic information for scene understanding. However, indoor construction workplaces are extremely low-texture and dynamic environments, with limited and stable features for visual SLAM. Although structural SLAM can extract geometric features for localization, most of these methods rely on point and line features, which are sparse in indoor construction workplaces, in which there are only a few planes such as walls, ceilings and floors, without interior decoration providing line features. The challenge is complicated further by the dynamic environments with moving workers and machines. How to extract reliable features for visual SLAM in such environments is a critical problem for indoor construction robots that remains under-investigated. Based on the observation that planes are the dominant geometric features in indoor construction workplaces, this paper proposes a plane-informed SLAM methodology to address the challenge by leveraging accurate plane detection and plane-based static region detection. Experiments on real-world construction sites indicated consistent improvement of over 50% in localization accuracy and demonstrate stable performance across different dynamic sites, outperforming other visual SLAM systems. This work demonstrates that planes can serve as the primary structural information and can contribute to a visual SLAM system robust to low-texture and dynamic indoor construction environments.

Journal of Computing in Civil EngineeringVol. 41(2)
Hong Kong University of Science and Technology (HK)
Robotics and Sensor-Based Localization
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon