Dynamic Map Update of Non-static Facility Logistics Environment with a Multi-robot System

Abstract

Autonomous robots need to perceive and represent their environments and act accordingly. Using simultaneous localization and mapping (SLAM) methods, robots can build maps of the environment which are efficient for localization and path planning as long as the environment remains unchanged. However, facility logistics environments are not static because pallets and other obstacles are stored temporarily. This paper proposes a novel solution for updating maps of changing environments (i.e. environments with low-dynamic or semi-static objects) in real-time with multiple robots. Each robot is equipped with a laser range sensor and runs localization to estimate its position. Each robot senses the change in the environment with respect to a current map, initially built with a SLAM method, and constructs a temporary map which will be merged into the current map using localization information and line features of the map. This procedure enables the creation of long-term mapping robot systems for facility logistics.