Lidar mapping improves with submap anchoring for lifelong accuracy
SEAM: Submap-Anchored Evidence for Lifelong LiDAR Mapping under Trajectory Deformation
Robotics
Summary
Mapping a place with LiDAR is tricky when you revisit or change parts later. The authors created SEAM, which breaks a big map into smaller sections called submaps to keep track of changes more reliably. This method handles moving objects and environmental changes better, even when the map’s path is updated. It helps make mapping faster and more accurate over time at places like construction sites.
What this means in practice
- •For construction site managers: Maintain accurate 3D site maps over multiple sessions despite changes in equipment and environment.
- •For autonomous vehicle teams: Improve long-term environmental mapping by reducing errors from unreliable loop closures and dynamic objects.
Authors
Kyuwon Kim
Abstract
We propose SEAM, a LiDAR-based lifelong mapping framework. Instead of relying on a single anchor spanning the entire session, SEAM generates evidence based on a trajectory optimized with submap-level anchors, and performs dynamic object removal and change detection. Through submap-level reprojection, the generated evidence remains usable even if the trajectory is subsequently modified by a new session, eliminating the need to recompute the entire process from scratch. SEAM suppresses geometrically unreliable inter-session loop edges using a DOP-based confidence measure. Suppressing unreliable loop edges prevents alignment errors. SEAM also uses a directional voxel-wise evidence model. The model accounts for occupancy patterns that vary with ray direction. Direction-aware evidence separates dynamic objects from environmental changes more precisely. Experiments on a real construction-site dataset and a long-term multi-session dataset show that SEAM achieves higher accuracy and faster processing than existing methods.