RBLAM — Reflector-Based Localization
Sole inventor of production-grade reflector-based localization and mapping, deployed at industrial customer sites.
Problem
Large industrial facilities (warehouses, manufacturing plants) pose severe challenges for visual and LiDAR SLAM — repetitive structure, scene dynamism, and map scale cause localization failures in GPS-denied natural navigation systems.
Approach
I invented RBLAM (Reflector-Based Localization and Mapping) and led its production-grade development:
- Sparse reflector-landmark representation with probabilistic data association
- Integration into the production navigation stack with online map management
- Fail-over logic for continuous operation when primary localization modalities degrade
- Related patent filed — A Mobile Entity Localization System and a Method Thereof
Results
- Sole inventor and technical lead for algorithm and deployment
- Deployed at 2+ customer sites in industrial environments
- Enables robust relocalization where feature-based methods struggle
- Complements natural navigation with a dedicated sparse-landmark modality
Technologies
ROS 2 · C++ · GTSAM · LiDAR · Probabilistic estimation · Map management