Abstract The paper presents a method of determination an accurate position of a target (rover, immobile sensor, astronaut) on surface of the Moon or other celestial body devoid of navigation infrastructure (like Global Positioning System), by using a group of self-calibrating rovers, which serves as mobile reference points. The rovers are equipped with low-precision clocks synchronized by external broadcasting signal, to measure the moments of receiving radio signals sent by localized target. Based on the registered times, distances between transmitter and receivers installed on beacons are calculated. Each rover determines and corrects its own absolute position and orientation by using odometry navigation and measurements of relative distances and angles to other mobile reference points. Accuracy of navigation has been improved by the use of a calibration algorithm based on the extended Kalman filter, which uses internal encoder readings as inputs and relative measurements of distances and orientations between beacons as feedback information. The key idea in obtaining reliable values of absolute position and orientation of beacons is to first calibrate one of the rovers, using the remaining ones as reference points and then allow the whole group to move together and calibrate all the rovers in-motion. We consider a number of cases, in which basic modeling parameters such as terrain roughness, formation size and shape as well as availability of distance and angle measurements are varied.