SLAM and map merging

Ángel León García, Rafael Barea Navarro, Luis Miguel Bergasa Pascual, Elena López Guillén, Manuel Ocaña Miguel, David Schleicher Gómez


This paper presents a multi-robot mapping and localization system. Learning maps and efficient exploration of an unknown environment is a fundamental problem in mobile robotics usually called SLAM (simultaneous localization and mapping problem). Our approach involves a team of mobile robots that uses Scan-Matching and Fast-SLAM techniques based on scan-laser and odometry information for mapping large environments. The single maps generated for each robot are more accurate than the maps generated only by odometry. When a robot detects another, it sends its processed map and the master robot generates a very accurate global map. This method cuts down the global map building time. Some experimental results and conclusions are presented.


Multi-robot SLAM; Scan-matching; Fast-slam; Rao-blackwellised particle filter