A Line Of Sight approach for Multi-agent SLAM

Abstract

This paper presents an algorithm for Simultaneous Localization and Mapping (SLAM) by a multi agent system, for the spatial and planar cases. At each time step the agents’ positions are calculated relative to a single or a set of landmarks. Alternatively, positions can be calculated relative to agents with known positions. Each pair of agents which are in a line-of-sight-position helps the mapping task by "coloring" a segment in the obstacle-free-workspace in addition to the agents’ trajectories. Here, unlike traditional SLAM, the localization procedure is independent to the mapping. We present simulation results for the SLAM and show the advantage of using multi agent system for such tasks. In addition, we exemplify our approach using low-cost-opticsensors, placed on an omni wheeled platform and show the positioning accuracy.