AGLoc++: WiFi-Fused Global Localization and Monte Carlo Enhanced Tracking in Hierarchical Area Graph

Project Overview

AGLoc (link) proposes a robust indoor localization method using 3D LiDAR and hierarchical topo-semantic Area Graphs, achieving long-term stable performance in office environments by filtering dynamic clutter and matching architectural features (e.g., walls/doors), outperforming traditional SLAM approaches. This Project enhance the previous work in the following aspects:

1. Port from ROS1 to ROS2 (Done)
2. Implement WiFi-aided Kidnap Recovery (Done)
3. Integrated with Nav2 stack by replacing AMCL (Done)
4. indoor cross-level localizaiton (Done)
5. Implement Odom-fused Monte Carlo Tracking (Doing)
6. Re-localize when losing tracking (Doing)

Key Technologies

1. Long-term LiDAR localization framework based on the hierarchical Area Graph, enabling robust global and local localization in dynamic indoor environments.
2. Clutter-adaptive subsampling to filter transient objects (e.g., furniture, pedestrians) from 3D LiDAR point clouds, preserving structural features (walls, doors).
3. Hypothesis-scoring global localization to address the kidnapped robot problem: Candidate poses are sampled, ranked via a novel Area Graph match metric, and refined for reliability.
4. Weighted point-to-line ICP with a clutter-aware weight function, ensuring pose tracking relies solely on lifelong architectural features (e.g., walls, passages).
5. Corridorness-aware downsampling to optimize point cloud registration in corridor-dominated spaces, improving ICP accuracy.

System Architecture

Future Work

Future development will focus on:

1. Implement Odom-fused Monte Carlo Tracking
2. Re-localize when losing tracking
3. indoor cross-level re-localization

Project Team

• Current Work: Jiajie Zhang (zhangjj2023@shanghaitech.edu.cn)
• Previous Work: Fujing Xie (xiefj@shanghaitech.edu.cn)