Revolutionizing electric robot charging infrastructure through federated transfer learning and data route optimization

The growing prevalence of electric robots within industrial settings calls for innovative solutions to address the limitations in the existing charging infrastructure, ultimately improving the efficiency and sustainability of these robotic systems. However, internet connectivity, which may be required for prior scheduling and charging, is often unavailable due to security concerns, harsh conditions, operational control requirements, and regulatory compliance. Protecting sensitive data, ensuring reliability, maintaining production control, and adhering to industry-specific regulations drive the decision to isolate internal networks from the internet. In such scenarios, locating a suitable charging spot for ERs, particularly during urgent situations with critically low battery levels, becomes challenging. This paper proposes a scheduling algorithm that relies on an ER-to-ER mechanism for charging at available stations. Moreover, since the process relies on ad-hoc communication susceptible to potential malicious actions by intermediate entities, it poses a risk to the overall process. To mitigate this risk, we propose a machine learning-driven method to detect malicious connections, safeguarding the network’s stability. Moreover, to reduce network congestion resulting from routing requests, we have modified the Grey-wolf optimization algorithm specifically for industrial settings. This modification enables the efficient scheduling of charging requests at optimal stations, all while prioritizing the privacy of robots.