Multirobot Collaborative Monocular SLAM Utilizing Rendezvous

Multi-robot teams can build a shared map faster than any single robot, but fusing each robot’s map usually calls for extra sensing to register them against one another. This work proposes a collaborative monocular SLAM system, including a map-fusion algorithm, that instead exploits rendezvous — moments when two team members come within view of each other during normal operation — using only a monocular camera on each robot, with no additional sensors required.

Received the 27th Samsung Human-Tech Paper Award (Silver Prize). Published in IEEE Transactions on Robotics, 2021.