TreeLoc++: Robust 6-DoF LiDAR Localization in Forests with a Compact Digital Forest Inventory

TL;DR

TreeLoc++ is proposed, a global localization framework that operates directly on DFIs as a discriminative representation, eliminating the need to use the raw point clouds, and achieves precise localization with centimeter-level accuracy.

Abstract

Reliable localization is essential for sustainable forest management, as it allows robots or sensor systems to revisit and monitor the status of individual trees over long periods. In modern forestry, this management is structured around Digital Forest Inventories (DFIs), which encode stems using compact geometric attributes rather than raw data. Despite their central role, DFIs have been overlooked in localization research, and most methods still rely on dense gigabyte-sized point clouds that are costly to store and maintain. To improve upon this, we propose TreeLoc++, a global localization framework that operates directly on DFIs as a discriminative representation, eliminating the need to use the raw point clouds. TreeLoc++ reduces false matches in structurally ambiguous forests and improves the reliability of full 6-DoF pose estimation. It augments coarse retrieval with a pairwise distance histogram that encodes local tree-layout context, subsequently refining candidates via DBH-based filtering and yaw-consistent inlier selection to further reduce mismatches. Furthermore, a constrained optimization leveraging tree geometry jointly estimates roll, pitch, and height, enhancing pose stability and enabling accurate localization without reliance on dense 3D point cloud data. Evaluations on 27 sequences recorded in forests across three datasets and four countries show that TreeLoc++ achieves precise localization with centimeter-level accuracy. We further demonstrate robustness to long-term change by localizing data recorded in 2025 against inventories built from 2023 data, spanning a two-year interval. The system represents 15 sessions spanning 7.98 km of trajectories using only 250KB of map data and outperforms both hand-crafted and learning-based baselines that rely on point cloud maps. This demonstrates the scalability of TreeLoc++ for long-term deployment.

Figures (27)

• Figure 1: (Left) Images captured from the same location in the Evo forest two years apart, showing vegetation growth (yellow) and denser foliage (cyan), which amplifies perceptual aliasing. (Right) TreeLoc++ achieves accurate multi-session alignment using only lightweight DFIs by leveraging tree attributes, avoiding raw point clouds that can vary across sessions due to different LiDAR sensors (black).
• Figure 2: Pipeline of TreeLoc++. TreeLoc++ extracts tree-level traits using RealtimeTrees and converts them into two descriptors for place recognition. Based on the retrieved candidates, it estimates the full 6-DoF pose while suppressing outliers arising from incorrect triangle matches. Using the resulting global localization, trees observed across multiple sessions can be associated, enabling consistent matching and incremental updates of tree traits.
• Figure 3: Payload-based forest inventory generation. (Top) Payloads are aggregated over temporal windows to form local inventories along the trajectory. (Bottom) All payloads are merged into a global inventory, which is queried at sampled poses to generate local inventories for localization.
• Figure 4: Comparison of alignment strategies for 2D projection. Arrows indicate the estimated vertical direction. (a) Raw point clouds show inconsistent orientation. (b) Ground-based alignment is inconsistent across views. (c) Axis-based alignment using tree stems yields consistent orientation.
• Figure 5: Generation of histogram descriptors. (a) TDH encodes tree distributions using overlapping radial and DBH bins. Each radial bin spans from a dashed circle to the second subsequent solid circle, illustrating the overlapping construction. (b) PDH aggregates inter-tree distance histograms computed for each tree into a 1D scene-level representation.