This repository is the official implementation of the paper:
Collaborative Dynamic 3D Scene Graphs for Open-Vocabulary Urban Scene Understanding
Tim Steinke*, Martin Büchner*, Niclas Vödisch*, and Abhinav Valada.
*Equal contribution.arXiv preprint arXiv:2503.08474, 2025
If you find our work useful, please consider citing our paper:
@article{steinke2025curbosg,
author={Steinke, Tim and Büchner, Martin and Vödisch, Niclas and Valada, Abhinav},
title={Collaborative Dynamic 3D Scene Graphs for Open-Vocabulary Urban Scene Understanding},
journal={arXiv preprint arXiv:2503.08474},
year={2025},
}
Make sure to also check out our previous work on this topic: CURB-SG.
Mapping and scene representation are fundamental to reliable planning and navigation in mobile robots. While purely geometric maps using voxel grids allow for general navigation, obtaining up-to-date spatial and semantically rich representations that scale to dynamic large-scale environments remains challenging. In this work, we present CURB-OSG, an open-vocabulary dynamic 3D scene graph engine that generates hierarchical decompositions of urban driving scenes via multi-agent collaboration. By fusing the camera and LiDAR observations from multiple perceiving agents with unknown initial poses, our approach generates more accurate maps compared to a single agent while constructing a unified open-vocabulary semantic hierarchy of the scene. Unlike previous methods that rely on ground truth agent poses or are evaluated purely in simulation, CURB-OSG alleviates these constraints. We evaluate the capabilities of CURB-OSG on real-world multi-agent sensor data obtained from multiple sessions of the Oxford Radar RobotCar dataset. We demonstrate improved mapping and object prediction accuracy through multi-agent collaboration as well as evaluate the environment partitioning capabilities of the proposed approach.
We will release the code upon the acceptance of our paper.
We separately release our developed tool for multi-agent urban mapping based on the Oxford Radar RobotCar Dataset. Please find the code at this link: https://github.com/TimSteinke/multi_robotcar
For academic usage, the code is released under the GPLv3 license. For any commercial purpose, please contact the authors.
We thank Kenji Koide for open-sourcing the ROS package hdl_graph_slam that we use as base for our multi-agent LiDAR SLAM framework.
This work was funded by the German Research Foundation (DFG) Emmy Noether Program grant number 468878300.
