Researchers at Karlsruhe Institute of Technology (KIT) have developed a federated learning method for training robots without sharing sensitive data. The method allows robots from different companies and locations to learn from each other without central data collection. The project, called FLAIROP (Federated Learning for Robot Picking), involved partners from Canada and Germany and focused on object recognition, deep learning, and data security. The researchers are now seeking industry and research partners to further develop the federated learning system.
Source: Link