II-NVM: Enhancing Map Accuracy and Consistency with Normal Vector-Assisted Mapping

TL;DR

II-NVM tackles indoor SLAM double-sided mapping by exploiting normal vector consistency within an enhanced voxel map. It extends voxel blocks to store front and back normals, uses a distance-adaptive neighborhood (adaptive radius KD-tree) for robust normal estimation, and employs an LRU cache to enable efficient incremental updates. Data association leverages normal-vector–aware constraints to distinguish front/back surfaces, improving map accuracy and trajectory estimation. Validated in Gazebo simulations and real indoor data, II-NVM achieves lower ATE and accurate wall-thickness estimation, and the authors open-sourced code and a dedicated dataset for double-sided mapping.

Abstract

SLAM technology plays a crucial role in indoor mapping and localization. A common challenge in indoor environments is the "double-sided mapping issue", where closely positioned walls, doors, and other surfaces are mistakenly identified as a single plane, significantly hindering map accuracy and consistency. To address this issue this paper introduces a SLAM approach that ensures accurate mapping using normal vector consistency. We enhance the voxel map structure to store both point cloud data and normal vector information, enabling the system to evaluate consistency during nearest neighbor searches and map updates. This process distinguishes between the front and back sides of surfaces, preventing incorrect point-to-plane constraints. Moreover, we implement an adaptive radius KD-tree search method that dynamically adjusts the search radius based on the local density of the point cloud, thereby enhancing the accuracy of normal vector calculations. To further improve realtime performance and storage efficiency, we incorporate a Least Recently Used (LRU) cache strategy, which facilitates efficient incremental updates of the voxel map. The code is released as open-source and validated in both simulated environments and real indoor scenarios. Experimental results demonstrate that this approach effectively resolves the "double-sided mapping issue" and significantly improves mapping precision. Additionally, we have developed and open-sourced the first simulation and real world dataset specifically tailored for the "double-sided mapping issue".

Figures (7)

• Figure 1: illustrates a real-world scenario where the double-sided mapping issue arises, presenting examples of both erroneous and correct mappings.
• Figure 2: System overview of II-NVM.
• Figure 3: When the LiDAR scans through a wall, two closely spaced walls may be erroneously identified as a single plane. Accurate matching is achieved by employing normal vector consistency calculations.
• Figure 4: Gazebo simulation scenarios show (a) garage scene, (b) café scene, and (c) fixed-distance wall scene.
• Figure 5: Comparison of localization estimates from different algorithms.