Model-Based Relative Localization for Cooperative Robots Using Stereo Vision

In the last years, Multi-Robot Systems (MRS) have been receiving great attention, as they can be effectively employed in several fields. Generally, for a collaborative behaviour to be successful, a precise localization strategy is required. A number of collective positioning schemes are available in literature, which mainly differ depending on the sensors and on the cooperation strategies adopted. In this work, we propose a model-based relative localization method using stereo vision, which enables a complex agent, equipped with a stereo head, to simultaneously detect and localize several small robots, navigating in a coordinated manner for a common task. The paper describes the method in detail and presents experimental tests performed on a real multi-agent system, proving the method to be accurate and effective for multi-robot localization, and environment exploration and mapping.