LiDAR Odometry for Inland Waterways using Water-Plane Constraints

Iulian Filip, Jaime Godoy, Daniel Medina

Keywords: LiDAR SLAM, Inland Waterways, Water Plane Detection

TL;DR: We present a LiDAR odometry algorithm optimized for inland waterway environments by using water surface detection as a constraint.

Abstract: LiDAR odometry is a key component of autonomous navigation, yet methods that perform well on roads and indoors often degrade in inland waterways (IWs) due to sparse and noisy water-surface returns, long-range observations, and vessel motion induced by waves and currents. We propose a lightweight LiDAR-only odometry pipeline that leverages the water surface as a domain-specific geometric prior. The approach propagates the pose with a constant-velocity motion model and fuses two complementary measurement cues within an error-state Kalman filter: (i) a water-plane observation, obtained from water-surface detection and parameterized by its normal and distance, which constrains elevation and tilt, and (ii) a scan-to-submap ICP update that refines the full 6-DoF state. We evaluate on a self-collected IW dataset spanning 14.5 km over 103 min, covering vegetation corridors, dense urban sections, port environment, narrow channels and bridge underpasses, with synchronized measurements from three LiDARs: Velodyne VLP-32, Ouster OS0-128, and Livox Avia. Compared to the state-of-the-art LiDAR-only baseline KISS-ICP, our method consistently reduces trajectory error across scenes and sensors, achieving more than 80\% lower absolute position error and more than 60\% lower absolute orientation error.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Paper Acceptance: No

Submission Number: 5