Multi-robot mapping for optically guided vehicles

This paper addresses the problem of mapping and localization for optically guided Automated Guided Vehicles (AGV) that are widely needed in the warehousing industry. It describes an approach to build a floor plan of the optical path traversed by the AGV, while keeping the process cost-effective by using the AGVs internal sensors and the optical sensor used to keep track of the path. The idea is to create a distributed system where each AGV knows its location in the warehouse and is capable of calculating an optimal path when given a destination without having to map the environment prior. This is done through fusion of maps from multiple AGV. Each AGV creates its own map and communicates it to the mapping module which collects maps from all the AGV in the system and creates a fused map. Such a map needs to be robust against errors and noise from multiple AGV.