This work implements and extends the TripletLoc semantic triplet global localization framework (Ma et al., IEEE RA-L 2025).

@ARTICLE{Ma2025TripletLoc,
  author={Ma, Weixin and Yin, Huan and Wong, Patricia J. Y. and Wang, Danwei and Sun, Yuxiang and Su, Zhongqing},
  journal={IEEE Robotics and Automation Letters},
  title={TripletLoc: One-Shot Global Localization Using Semantic Triplet in Urban Environments},
  year={2025},
  volume={10},
  number={2},
  pages={1569-1576}
}

Multi-scale: In config/config_tripletloc.yaml, multi-scale settings are configured jointly by multi_scale_edge_dis_thr and multi_scale_dis_resolution, which define the adjacency rings per scale and the edge-length discretization step; mcq_min_inliers sets the minimum inlier count for maximum-clique matching at each scale (one entry per scale, aligned with the above).

Please refer to the following link for the M2Loc video: https://youtu.be/57yC7DsG6AU

---

If you find this paper helpful, please consider citing:

@INPROCEEDINGS{Lin2025ROBIO,
  author={Lin, Shuyue and Li, Jinsong and Sun, Yuxiang},
  booktitle={2025 IEEE International Conference on Robotics and Biomimetics (ROBIO)},
  title={Multi-Scale Triplet Descriptors for Global LiDAR Localization with Maximum Clique-Based Enhancement},
  year={2025},
  volume={},
  number={},
  pages={303-308}}