Federated Learning for Wireless Communication Prediction - The Use Case of Internal Logistic System Based on AGV

Automated Guided Vehicles (AGVs) are key components of internal logistics systems used for smart manufacturing. The shift from mass production to flexible, short-series production demands frequent and rapid adaptation of AGV routes to individualized logistics tasks. One of the key challenges is to ensure stable and reliable communication with AGVs in a dynamically changing industrial environment. To face this challenge we propose a novel approach for wireless communication parameters prediction based on federated learning paradigm. The communication parameters collected from AGVs' fleet during standard logistics operations are combined with data from navigation system and other key AGV's parameters including battery condition. Such approach allows for precise prediction of key communication parameters throughout the entire path planned for given AGV. Proposed solution utilizes a mas of data available in existing AGVs' eco system in order to minimize the number of additional wireless communication tests performed in the real production system, reduces efforts necessary to implement new logistic tasks, and makes communication more reliable.