A new paper on arXiv introduces SGM-SLAM, a distributed simultaneous localization and mapping framework for teams of robots operating where GPS is unavailable and communication is limited. The approach is built around scene graph matching, which the title suggests is key to making the system data-efficient.

The framework is explicitly aimed at multi-robot operation in challenging conditions, where bandwidth and connectivity are scarce. By focusing on data efficiency, it seeks to let robots share only the most relevant information rather than raw sensor streams.

Details beyond the abstract are not provided in the source, so the exact mechanism and performance claims remain to be seen. The paper is listed as a replacement, indicating revisions, but the abstract only gives a high-level description.