# SGBA: Semantic Gaussian Mixture Model-Based LiDAR Bundle Adjustment

arXiv: https://arxiv.org/abs/2410.01618 · HTML: https://mlsanity.com/paper/2410.01618/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 17 of its 32 references that are in the map.
- First posted on arXiv 2 Oct 2024.

## Builds on (most cited first, 17 of 17)

- [Searching Efficient 3D Architectures with Sparse Point-Voxel Convolution](https://mlsanity.com/paper/2007.16100/) (2020)
- [FAST-LIO2: Fast Direct LiDAR-inertial Odometry](https://mlsanity.com/paper/2107.06829/) (2021)
- [SAM3D: Segment Anything in 3D Scenes](https://mlsanity.com/paper/2306.03908/) (2023)
- [KISS-ICP: In Defense of Point-to-Point ICP -- Simple, Accurate, and Robust Registration If Done the Right Way](https://mlsanity.com/paper/2209.15397/) (2022)
- [SuMa++: Efficient LiDAR-based Semantic SLAM](https://mlsanity.com/paper/2105.11320/) (2021)
- [BALM: Bundle Adjustment for Lidar Mapping](https://mlsanity.com/paper/2010.08215/) (2020)
- [X-ICP: Localizability-Aware LiDAR Registration for Robust Localization in Extreme Environments](https://mlsanity.com/paper/2211.16335/) (2022)
- [Multi-Modal Continual Test-Time Adaptation for 3D Semantic Segmentation](https://mlsanity.com/paper/2303.10457/) (2023)
- [Joint Alignment of Multiple Point Sets with Batch and Incremental Expectation-Maximization](https://mlsanity.com/paper/1609.01466/) (2016)
- [Probabilistic Data Association via Mixture Models for Robust Semantic SLAM](https://mlsanity.com/paper/1909.11213/) (2019)
- [SA-LOAM: Semantic-aided LiDAR SLAM with Loop Closure](https://mlsanity.com/paper/2106.11516/) (2021)
- [Efficient and Consistent Bundle Adjustment on Lidar Point Clouds](https://mlsanity.com/paper/2209.08854/) (2022)
- [Rigid and Articulated Point Registration with Expectation Conditional Maximization](https://mlsanity.com/paper/2012.05191/) (2020)
- [SLICT: Multi-input Multi-scale Surfel-Based Lidar-Inertial Continuous-Time Odometry and Mapping](https://mlsanity.com/paper/2211.03900/) (2022)
- [Outram: One-shot Global Localization via Triangulated Scene Graph and Global Outlier Pruning](https://mlsanity.com/paper/2309.08914/) (2023)
- [On Bundle Adjustment for Multiview PointCloud Registration](https://mlsanity.com/paper/2108.02976/) (2021)
- [PSS-BA: LiDAR Bundle Adjustment with Progressive Spatial Smoothing](https://mlsanity.com/paper/2403.06124/) (2024)

## Built on by (most cited first, 1 of 1)

- [NVP-HRI: Zero Shot Natural Voice and Posture-based Human-Robot Interaction via Large Language Model](https://mlsanity.com/paper/2503.09335/) (2025)

## Abstract (arXiv)

LiDAR bundle adjustment (BA) is an effective approach to reduce the drifts in pose estimation from the front-end. Existing works on LiDAR BA usually rely on predefined geometric features for landmark representation. This reliance restricts generalizability, as the system will inevitably deteriorate in environments where these specific features are absent. To address this issue, we propose SGBA, a LiDAR BA scheme that models the environment as a semantic Gaussian mixture model (GMM) without predefined feature types. This approach encodes both geometric and semantic information, offering a comprehensive and general representation adaptable to various environments. Additionally, to limit computational complexity while ensuring generalizability, we propose an adaptive semantic selection framework that selects the most informative semantic clusters for optimization by evaluating the condition number of the cost function. Lastly, we introduce a probabilistic feature association scheme that considers the entire probability density of assignments, which can manage uncertainties in measurement and initial pose estimation. We have conducted various experiments and the results demonstrate that SGBA can achieve accurate and robust pose refinement even in challenging scenarios with low-quality initial pose estimation and limited geometric features. We plan to open-source the work for the benefit of the community https://github.com/Ji1Xinyu/SGBA.