Performance enhancement for outdoor cooperative localization considering of error correlation between GPS data

This paper presents an effective scheme to improve position estimation accuracy for multiple robot localization under outdoors. This scheme is applicable to cooperative localization (CL) system in which a robot exchanges no more relative position information between any of the robots. Applying the relative distance acquired by considering the error correlation of Global Positioning System (GPS) to cooperative localization framework, we can increase the positioning accuracy more than the CL system under certain circumstances without detecting other robot in a group. We consider a group of three robots equipped proprioceptive sensors such as odometer and exteroceptive sensors such as GPS and digital magnetic compass (DMC). In addition, each robot in the group is able to share the position data from sensors with others through wireless communication device. We prove to be strong error correlation between two robots' GPS position data acquired from experiment. Finally, in order to show the validity of our scheme, the proposed algorithm is implemented for multi-robot localization using experimental data of GPS error correlation.