ABSTRACT
Multi-mobile robot system can be used in collaboration for exploring, monitoring, rescuing, and so on. Localization is estimating the robot's pose. The simplest localization method is dead reckoning, where using the measurement data from the odometers mounted on the robots, the pose of each robot is independently estimated. Cooperative localization (CL) is a situation where a single robot uses other robots as dynamic landmarks. Compared with estimated alone, because of the additional information available, CL can achieve more reliable and precise localization than one robot alone. Graph-based CL is a batch processing approach where all the history poses are continuously updated along with increasing the new measurement data. CL can be implemented in the centralized- or distributed-architecture. In the centralized CL, there exists a processing center to execute the CL algorithm. The communication capacity is proportional to the equipment power. When heavy communication equipment is employed, it is in conflict with the finite payload capacity.
