Multi-robot simultaneous localization and uncertainty reduction on maps (MR-SLURM)

This paper presents two strategies for simultaneous localization and uncertainty reduction on maps for a team of robots. The proposed strategies differentiate between homogeneous and heterogeneous multi-robot teams assigning different roles based on risk and/or capabilities. We apply the proposed algorithms to the Robot-Camera Sensor Network localization problem, where a team of robots moves through an environment equipped with a camera sensor network. Each robot uses its own pose estimate to localize every camera it encounters. This is achieved by using the camera observation of the robot to extract a 6-DoF transformation between the camera and the robot. The inverted transformation places the camera in the robots global frame of reference, and map merging among the multiple robots places the cameras and the team of robots in a common frame of reference. At the core of the estimation, an extended Kalman filter algorithm is used to estimate the joint pose of robots and cameras. Experimental results from realistic simulations are presented that validate the proposed strategies.