Answers are grounded in this paper and cite their sources.
AI agents driven by foundation models have shown potential for industrial automation beyond rule-based control. However, existing studies differ in how industrial agents collaborate and how their systems are tested, making it difficult to compare networking requirements and physical execution capabilities. This paper examines the intersection of networked agents and industrial automation, where industrial signals, communication constraints, contextual information, and operational requirements must be considered jointly. It develops the Industrial Internet of Agents (IIoA) and presents a conceptual architecture for analyzing industrial deployment requirements. Building on the Internet of Agents (IoA) framework, IIoA re-specifies sensing, communication, decision-making, and execution through eleven industrial extensions. Representative studies are synthesized across shopfloor operations, autonomous plant operations and maintenance, supply chain collaboration, and digital twin systems. The analysis indicates that industrial agent networking requires the joint design of semantic coordination, operational technology (OT) interoperability, latency-aware deployment, trust governance, and verifiable physical execution. Finally, seven open challenges are identified, covering reliability, latency, protocol integrity, scalability, physical grounding, adaptability, and evaluation.
No takes yet. Share an insight, caveat, or question.
Li et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: