Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces a novel framework that utilizes camouflage-matched fiducial markers made from Cholesteric Spherical Reflectors (CSRs) to georeference LiDAR-odometry trajectories and dense RTAB-Map reconstructions in GNSS-denied environments. By employing a combination of coarse similarity alignment and marker-constrained pose-graph optimization, the method achieves significant reductions in revisit inconsistency—97.9% for UAV and 99.1% for UGV sessions—while maintaining real-time processing capabilities. The results highlight the potential for accurate spatial referencing without GNSS, paving the way for improved autonomous operations in challenging environments.
Marker-constrained pose-graph optimization slashes revisit inconsistency by up to 99.1%, enabling reliable georeferencing in GNSS-denied settings.
Autonomous operation in GNSS-denied environments requires heterogeneous mapping pipelines to maintain a consistent spatial reference. This paper presents a framework using camouflage-matched fiducial markers fabricated from Cholesteric Spherical Reflectors (CSRs) as pre-surveyed visual anchors. The anchors georeference both a lightweight LiDAR-odometry trajectory and a dense RTAB-Map reconstruction, allowing their outputs to be expressed in a common LUREF frame (geodetic coordinate reference system used in Luxembourg) without requiring GNSS measurements during operation. The method combines coarse similarity alignment with marker-constrained pose-graph optimization. We evaluate it using two handheld acquisition sessions with ground-level and elevated motion profiles emulating UGV and UAV operation. A single iMarker was relocated among six surveyed positions, with the first position revisited to quantify drift correction. Marker-anchor correction reduced revisit inconsistency by 97.9% and 99.1% for the UAV- and UGV-emulating sessions, respectively, and improved held-out anchor prediction compared with one-time alignment. Separately georeferenced dense reconstructions achieved a median cross-session nearest-neighbour distance of 58 cm without explicit cross-session registration. Marker processing operated in real time, while trajectory correction required less than 0.25 s per session. These results demonstrate a proof of concept for georeferencing lightweight odometry and dense reconstructions using visually unobtrusive, pre-surveyed anchors during GNSS-denied operation.