Multi-agent distributed collaborative mapping provides comprehensive and efficient representations for robots. However, existing approaches lack instance-level awareness and semantic understanding of environments, limiting their effectiveness for downstream applications. To address this issue, we propose OpenMulti, an open-vocabulary instance-level multi-agent distributed implicit mapping framework. Specifically, we introduce a Cross-Agent Instance Alignment module, which constructs an Instance Collaborative Graph to ensure consistent instance understanding across agents. To alleviate the degradation of mapping accuracy due to the blind-zone optimization trap, we leverage Cross Rendering Supervision to enhance distributed learning of the scene. Experimental results show that OpenMulti outperforms related algorithms in both fine-grained geometric accuracy and zero-shot semantic accuracy. In addition, OpenMulti supports instance-level retrieval tasks, delivering semantic annotations for downstream applications. The project website of OpenMulti is publicly available at https://openmulti666.github.io/.
@article{arxiv.2509.01228,
title = {OpenMulti: Open-Vocabulary Instance-Level Multi-Agent Distributed Implicit Mapping},
author = {Jianyu Dou and Yinan Deng and Jiahui Wang and Xingsi Tang and Yi Yang and Yufeng Yue},
journal= {arXiv preprint arXiv:2509.01228},
year = {2025}
}
Comments
Accepted to IEEE Robotics and Automation Letters. Project website: https://openmulti666.github.io/