Federated Learning-Based Human-Robot Collaboration with Explainable AI: A Framework for Secure and Transparent Interaction

摘要

The advancement of Human-Robot Collaboration (HRC) is pivotal for achieving intelligent, autonomous, and safe cooperative systems across industries such as healthcare, manufacturing, and service robotics. However, traditional centralized learning methods pose significant risks related to data privacy, scalability, and adaptability. Moreover, opaque decision-making processes in robotic systems hinder human trust and effective collaboration. This paper introduces a novel framework, Federated Learning-Based Human-Robot Collaboration with Explainable AI (FL-HRC-XAI), aimed at enabling secure, transparent, and adaptive human-robot teaming.The proposed framework leverages Federated Learning (FL) to allow robots to learn collaboratively from decentralized data without sharing raw information, thus preserving user privacy and mitigating data exposure risks. To address the crucial need for trust and transparency, the system integrates Explainable AI (XAI) modules that generate personalized, real-time explanations for robot behaviors. The framework further incorporates robust aggregation methods and differential privacy mechanisms to defend against adversarial attacks and information leakage during the federated training process. We demonstrate that the FL-HRC-XAI framework achieves task performance comparable to centralized baselines while significantly enhancing user trust, satisfaction, and collaboration fluency. Overall, the research findings validate the practical viability of combining federated learning, explainable AI, and adaptive human-robot collaboration for secure, transparent, and effective human-robot teaming.