EdgePLC: AI, IT & OT Industrial Edge Controller

Most factories today still operate with three disconnected layers:

• OT (Operational Technology): PLCs, DCS, sensors, and actuators handle real-time control — stable but closed and rigid.
• IT (Information Technology):MES, ERP, SCADA, and cloud platforms manage data and business — powerful but far from the shop floor.
• AI (Artificial Intelligence):Vision models, predictive maintenance, and process optimization run in the cloud or on separate industrial computers — separated by gateways and protocol conversion.

The result: AI detects an anomaly, but the PLC has not received the command. MES issues a work order, but the line still runs the old recipe. Engineers must manually change PLC programs.

This is not smart manufacturing. It is three automation islands stitched together.

1.The EdgePLC Solution: Three Levels of Integration

The EdgePLC Industrial AI Edge Controller was designed to break down these walls.

1.1. Software-Hardware Decoupling
EdgePLC uses an SOM (System on Module) platform. Users can choose compute from Allwinner T113-i to Rockchip RK3588J — same host, same I/O system, scalable compute. It runs Ubuntu 20.04 + Linux-RT and natively supports CODESYS Runtime, OpenPLC, NexPLC, and IEC 61131-3 / IEC 61499. Existing PLC programs migrate directly. No vendor lock-in.

1.2. IT-OT Convergence
Control and data close the loop in one device. CODESYS Runtime runs on Cortex-A76/A55 cores; the NPU runs AI inference; Ubuntu runs Node-RED, Docker, and BLIoTLink. From sensor acquisition to AI inference to control execution to cloud upload — all in one device, with no cross-device latency or protocol loss.

1.3. AI Integrated into Control
AI moves from “seeing” to “doing”:

• Phase 1: AI visual safety and quality inspection — YOLOv5/8 on NPU, USB 3.1 cameras, N-series I/O directly drives alarm lights or reject cylinders.
• Phase 2: AI connects to MES and automatically adjusts recipes — reactor temperature, injection holding time, welding current — written directly into CODESYS logic.
• Phase 3: Continuous self-optimization — the system automatically adjusts detection thresholds and process parameters based on real-time feedback.

2.From Automation to Autonomy

Tomorrow’s autonomous line: MES orders “1,000 units of model A.” AI parses requirements, generates control parameters, and adjusts equipment. During production, AI monitors quality and optimizes parameters in real time. When the order switches to model B, AI recalculates and switches recipes — no manual intervention.

This is the natural evolution of AI + IT + OT convergence on EdgePLC.

3.Why EdgePLC?

• Open hardware:SOM-based, scalable compute, no single-chip lock-in.
• Open software:Ubuntu + CODESYS + OpenPLC + Node-RED + Docker, no single-software lock-in.
• Integrated control and compute: Real-time control, AI inference, protocol conversion, and cloud upload in one device — the physical foundation for AI + IT + OT convergence.

4.See EdgePLC at CIIF 2026

October 12–16, 2026, Shanghai National Exhibition and Convention Center.
CODESYS and Allwinner Technology will showcase EdgePLC at their booths.
Joint exhibition area 5.2HA168 will display EdgePLC solutions based on Allwinner T113-i / T507-H. Live demos available.

5.Conclusion

AI + IT + OT convergence is not addition — it is multiplication. When control, computing, and intelligence work together on one device, the value far exceeds three separate systems.

EdgePLC provides the open, flexible, unlocked hardware foundation for this shift.
From “three skins” to “three-in-one,” the next-generation smart manufacturing platform has arrived.