Event

CoMo3R-SLAM Collaborative Monocular Dense SLAM System Released

1 reports · 1 sources

Zhihao Cao and other authors proposed and released CoMo3R-SLAM, a collaborative monocular dense SLAM system for outdoor multi-agent systems. The system is centered on a learned feed-forward 3D reconstruction prior, uses dense point maps to anchor the scale across agents, and provides correspondences sufficient to geometrically verify inter-agent associations. According to the authors, the system requires neither measured depth nor pre-provided camera intrinsics; it runs at approximately 6–8 FPS on an RTX 3080 Ti. The authors state that its applicable scope also covers long-distance traversals, independently collected day and night video streams, and teams of up to four agents.