• AgentOptics framework uses agentic AI for autonomous optical system control via MCP. • Interprets natural language tasks, executes protocol-compliant actions across heterogeneous optical devices. • 64 MCP tools across 8 devices, 410-task benchmark tests understanding, coordination, robustness. • Achieves 87.7-99% task success, outperforms code-generation baselines (~50%). • Demonstrated in five case studies: DWDM provisioning, ARoF 5G fronthaul, polarization stabilization, DAS monitoring. • Establishes scalable, robust paradigm for autonomous control and orchestration of heterogeneous optical systems.

Article Summaries:

  • AgentOptics is an agentic AI framework that enables autonomous, high‑fidelity control of heterogeneous optical devices using the Model Context Protocol (MCP). The system translates natural‑language commands into protocol‑compliant actions via 64 standardized MCP tools spanning eight representative optical instruments, and evaluates performance on a 410‑task benchmark covering request understanding, role awareness, multi‑step coordination, linguistic robustness, and error handling. In tests, AgentOptics achieves 87.7 %-99.0 % task success, outperforming LLM‑based code‑generation baselines (up to 50 %). Five case studies demonstrate its applicability to system orchestration, monitoring, and closed‑loop optimization in DWDM, ARoF, 5G fronthaul, polarization stabilization, and distributed acoustic sensing.

Sources: