Solving Geometric Constraints for Relative Position Estimation using UWB Sensors in Multi-Robot System

Abstract

Recently, unmanned robots are replacing humans in hazardous areas. In particular, in order to perform complex tasks in a large-scale environment, collaboration between multiple robots is required rather than using a single robot. For this reason, a lot of research is being conducted on multi-robot systems. In a multi-robot system, each robot performs a task within a group of robots, so unlike a single-robot system, estimating the relative position with the other robots is as important as estimating the absolute position of a single robot in a single-robot system. One probable situation a multi-robot system faces is a GNSS-denied environment. Under GNSS-denied situations such as during combat or indoor navigation, a robot has to depend solely on the sensors mounted onto itself to achieve localization. In this paper, to determine the relative positions between all robots, an aggregated Euclidian distance matrix is employed. The matrix is a collection of relative distances measured by a UWB sensor on each robot. Then the relative positions of the robots are estimated using the distance information and geometric constraints. In addition to knowing the relative position between the robots, another significant advantage is that they can be used as initial guesses or constraints for faster localization.