Building and managing maps for autonomous mobile robots

International audiencePerceiving, representing and even understanding the environment that surrounds them is at the core of the autonomy of mobile robots. Environment representations are indeed required for any decision robots have to take, from simple obstacle avoidance motions to much more complex tasks such as cooperating to explore or patrol an area. The robotics research community has developed a whole corpus of approaches to endow mobile robots with the ability to represent their environment, using primarily the robots on-board sensors, but also various other available sources of information. The robot localization plays a central role in these developments, and has in particular lead to the development of wholesome estimation formalisms. The talk provides an overview of these methods, exhibits some links with Geographic Information Sciences, and sketches the current and future challenges in the area