Inland-LOAM: Voxel-Based Structural Semantic Mapping for Inland Waterways

Zhongbi Luo, Yunjia Wang, Jan Swevers, Peter Slaets, Herman Bruyninckx · 2025

Read Inland-LOAM: Voxel-Based Structural Semantic Mapping for Inland Waterways by Zhongbi Luo, Yunjia Wang, Jan Swevers, Peter Slaets, Herman Bruyninckx (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-03672.txt.

Introduction

Zhongbi Luo and co-authors present a mapping system for inland waterways in this 2025 paper. Boats need exact maps to move safely, but charts lack live detail and standard laser mapping fails near water. The system, called Inland-LOAM, uses laser scans to track a boat and build a map at the same time.

The method adds a check on the water surface to cut up-down drift. It turns 3D scans into flat maps that show features such as bridge height, using small blocks to judge shape. It also traces shorelines and saves them in a chart-friendly form. Tests on real data gave better place accuracy than other leading methods.

Catalog record

CREATORZhongbi Luo, Yunjia Wang, Jan Swevers, Peter Slaets, Herman Bruyninckx
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE64 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2508-03672

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Inland-LOAM: Voxel-Based Structural Semantic Mapping for Inland Waterways Zhongbi Luo, Yunjia Wang, Jan Swevers, Peter Slaets, Herman Bruyninckx Abstract Accurate and up-to-date geospatial information is crucial for enhancing the safety and autonomy of Inland Waterway Transport (IWT). Inland Electronic Navigational Charts (IENC) often lack detailed real-time information, conventional LiDAR-based Simultaneous Localization and Mapping (SLAM) methods have difficulty handling distinctive inland waterway characteristics, such as dynamic shorelines and challenging reflective water surfaces. These factors lead to significant localization drift, particularly in the vertical direction, and produce dense point cloud maps without the semantic richness required for effective autonomous decision making. This paper introduces Inland-LOAM, a comprehensive LiDAR-based SLAM framework specifically for inland waterway environments. Firstly, we present an improved feature extraction method adapted to the unique geometric characteristics of waterways, combined with a joint optimization framework that incorporates the water surface as a global planar constraint to significantly mitigate vertical drift. Secondly, we propose an innovative pipeline to transform dense 3D point cloud outputs into structured 2D semantic maps by constructing semantic voxel grids and performing geometric analyses (roughness, planarity, and slope). This enables classification into structurally meaningful categories and supports real-time computation of critical navigational parameters such as vertical bridge clearances. Additionally, an automated map conversion module efficiently extracts shoreline boundaries via ray-casting and exports these features into a lightweight, IENC-compatible format. Extensive evaluations were conducted using a diverse, real-world dataset featuring urban canals, vegetated banks, and bridge crossings. Comparative results demonstrate that Inland-LOAM achieves superior localization accuracy over state-of-the-art LiDAR SLAM methods. Generated semantic maps and extracted shoreline boundaries align with real-world conditions, as verified through comparisons with reference maps, thus providing reliable and up-to-date geospatial information to enhance navigational situational awareness. Both the dataset and the algorithm are publicly available to support future research. Upon publication, the code and datasets will be made publicly accessible. 1 Introduction IWT constitutes an essential component of Europe’s freight infrastructure, spanning a network exceeding 41,000 km, interlinking major cities and industrial hubs across 13 interconnected Member States [ 1 ] . As efforts increase to shift freight from congested road and rail networks, the importance of accurate geospatial information and detailed environmental models for managing and navigating these waterways grows [ 2 ] . However, navigating these waterways involves safety risks, with studies identifying human error as the primary cause in up to 70% of maritime incidents [ 3 ] . In this context, advancements in automation, especially the deployment of uncrewed surface vehicles (USVs), are driving a transformation in inland waterway transport. USVs play a crucial role not only by mitigating the risks associated with human factors, but also by promising improved cargo transport efficiency and facilitating innovative logistics services, including operations within challenging environments such as congested urban canal networks [ 4 ] . However, the effective operation of USVs is challenging due to the complex and dynamic nature of inland waterways. Safe and efficient navigation mainly depends on robust systems for perception, localization, and mapping. These systems must enable the vessel to reliably recognize diverse surroundings, including shorelines, infrastructure, vegetation, and overhead obstacles such as bridges, to operate safely. Traditionally, navigation on inland waterways relies highly on Global Navigation Satellite Systems (GNSS) for positioning, complemented by standard nautical charts like IENC for contextual spatial data [ 5 ] . However, GNSS frequently suffer from severe attenuation or complete outages in typical IWT scenarios, particularly during passages under bridges, within narrow urban canals flanked by tall buildings ’urban canyons’, or along waterways with dense vegetation canopies [ 6 , 7 ] . This documented unreliability highlights the need for onboard localization systems that work independently [ 8 ] . Furthermore, existing geospatial data products, IENC, while providing important routing and regulatory data, often lack the spatial detail or real-time updates required for accurate autonomous maneuvering and perception. Representations of shorelines and their surrounding features, including vegetation, shoreline constructions, and other obstacles, are often generalized or outdated in existing maps [ 9 , 10 ] . A key limitation of IENC is the representation of safety parameters, such as vertical bridge clearance, using static data. This approach fails to account for dynamic water level variations, a significant operational challenge in IWT affecting actual clearances [ 11 ] . As a result, relying on static, low-resolution geospatial information hinders the deployment of advanced autonomous functions. LiDAR-based SLAM has developed as a strong technique in terrestrial robotics and autonomous driving, making it possible to build precise 3D maps and accurate localization even in environments without GNSS [ 12 , 13 ] . Applying LiDAR SLAM to the IWT is a viable way to overcome the limitations of previous approaches. However, traditional LiDAR SLAM algorithms face particular challenges in this context. The main challenges come from the structure of the environment: unlike urban scenes with abundant planar facades (buildings) and consistent features (lane markings), inland waterways predominantly feature irregular, natural shorelines often covered with vegetation [ 11 ] . This lack of structured geometry and standardized visual cues affects the reliable extraction of edges and planar features necessary by algorithms like LOAM [ 14 ] and LeGO-LOAM [ 15 ] . In addition to these algorithmic challenges, the development and evaluation of SLAM algorithms for IWT is hindered by the lack of suitable benchmark datasets. Beyond achieving accurate geometric reconstruction, effective environmental understanding demands semantic understanding, e.g. distinguishing navigable water from vegetated banks, identifying a potential docking quay wall, or determining safe passage under a bridge. Standard SLAM methods typically do not provide this layer of structural semantic information. Finally, a gap exists in translating the rich, high-resolution data from SLAM systems into formats aligned with IENC navigation standards. Figure 1 : Our experimental platform, a catamaran named the Maverick, was equipped with a Sensor Box for situational awareness and data collection. To address these challenges, we propose Inland-LOAM, a novel voxel-based system designed for structural semantic mapping in inland waterway environments. This method achieves robust LiDAR-based localization and mapping adapted for water surfaces, extracts structural semantic information using geometric analysis within a voxel map, and generates enhanced environmental representations suitable for navigation and integration with standard chart formats. The contributions of this paper are as follows: 1. Inland-LOAM is specifically enhanced for robustness in GNSS-denied inland waterway environments through improved feature extraction performance and the incorporation of planar surface optimization. 2. A novel voxel-based structural semantic mapping approach using roughness, planarity, and slope analysis to generate 2D maps and calculating real-time bridge clearance. 3. An automated map conversion method to translate the high-detail structural semantic SLAM map into an IENC-compatible. 4. The proposed framework was demonstrated in a diverse set of real-world inland waterway scenes, including ports, urban canals, vegetated banks, bridges, and lifting gates. The experimental platform used for data collection is shown in Fig. 1 . All data are released as part of the publicly available dataset to support further research. 2 Related Work This section reviews existing literature relevant to LiDAR-based mapping and semantic understanding in the context of inland waterways, focusing on inland mapping methodologies, LiDAR odometry techniques, semantic scene understanding, and map integration with navigation standards. 2.1 Mapping in Inland Waterway Environments Inland waterways are mapped using various

Read the complete text