xiaojake / soslam

具有尺度比例和对称纹理约束的语义物体 SLAM-The implementation code for the paper SO-SLAM: Semantic Object SLAM with Scale Proportional and Symmetrical Texture Constraints.

SoSLAM

The implementation code for the paper SO-SLAM: Semantic Object SLAM with Scale Proportional and Symmetrical Texture Constraints, which has been submitted to RA-L&ICRA 2022.

Notes

We are preparing for the code and will release it soon in this page.

Abstract

Object SLAM introduces the concept of objects into Simultaneous Localization and Mapping (SLAM) and helps understand indoor scenes for mobile robots and object-level interactive applications. The state-of-art object SLAM systems face challenges such as partial observations, occlusions, unobservable problems, limiting the mapping accuracy and robustness. This paper proposes a novel monocular Semantic Object SLAM (SO-SLAM) system that addresses the introduction of object spatial constraints. We explore three representative spatial constraints, including scale proportional constraint, symmetrical texture constraint and plane supporting constraint. Based on these semantic constraints, we propose two new methods - a more robust object initialization method and an orientation fine optimization method. We have verified the performance of the algorithm on the public datasets and an author-recorded mobile robot dataset and achieved a significant improvement on mapping effects.

Citation

Liao, Z., Hu, Y., Zhang, J., Qi, X., Zhang, X. & Wang, W. (2021). SO-SLAM: Semantic Object SLAM with Scale Proportional and Symmetrical Texture Constraints.

Previous Work

[1] Liao, Z., Wang, W., Qi, X., & Zhang, X. (2020). RGB-D object SLAM using quadrics for indoor environments. Sensors, 20(18), 5150. [pdf]

[2] Liao, Z., Wang, W., Qi, X., Zhang, X., Xue, L., Jiao, J., & Wei, R. (2020). Object-oriented slam using quadrics and symmetry properties for indoor environments. Arxiv 2020. [pdf] [Code] [Video]