Why It Matters
Traditional PLCs are closed black boxes: hardware and software are tightly bound, and AI inference requires an external industrial PC. Industrial OS and AI agents need an open carrier that can host real-time control, edge computing, and AI inference together.
EdgePLC is designed from the chip level up to let control, computing, and inference share one system. The flagship BL245 uses Rockchip RK3588J with a built-in 6 TOPS NPU, runs YOLO-level vision models locally, executes CODESYS control logic deterministically on isolated cores, and drives physical I/O directly from inference results—closing the full chain inside one device.
Full Selection Matrix
|
Series |
Processor |
NPU |
Positioning |
Typical Use |
|
BL234 |
Allwinner T507-H (4×A53) |
— |
Entry-level control-display |
Basic logic control, HDMI 4K, logistics sorting |
|
BL237 |
RK3562J (4×A53+M0) |
1 TOPS |
Lightweight AI entry |
Visual inspection, AGV navigation |
|
BL241 |
RK3568J (4×A55) |
1 TOPS |
Lightweight AI enhanced |
Vision-guided robots, edge AIoT gateway |
|
BL244 |
RK3576J (4×A72+4×A53) |
6 TOPS |
Heavy-load AI |
Multi-camera vision, motion + AI fusion |
|
BL245 |
RK3588J (4×A76+4×A55) |
6 TOPS |
Flagship |
8K decoding, 1024-point I/O, complex industrial AI |
|
BL246 |
Raspberry Pi CM5 (BCM2712) |
Expandable Hailo |
Open-source ecosystem |
Raspberry Pi developers, AI accelerator expansion |
Three Industrial OS Adaptations
Beyond Three OS
CODESYS, vPLCnext, and WasomCodex are just the beginning. EdgePLC also supports OpenPLC, Ubuntu, and Docker, and can theoretically run any Linux-compatible industrial software. It is not closed hardware bound to one ecosystem, but a neutral foundation for OS developers, agent developers, and control engineers to build on the same device.
Traditional PLCs open only the upper garden while the底层 remains proprietary. EdgePLC pushes openness down to the operating system, runtime, and toolchain—giving users real choice over their technology stack.
FAQ
Q1: What exactly is EdgePLC?
A: EdgePLC is the world’s first open-source industrial hardware carrier launched, integrating an edge computing gateway, AI compute, and PLC control. It supports industrial OS such as CODESYS, vPLCnext, and WasomCodex.
Q2: What compute tiers does EdgePLC cover?
A: From entry-level BL234 (no NPU, control-display integration) to lightweight AI BL237/BL241 (1 TOPS NPU), to heavy-load AI BL244/BL245 (6 TOPS NPU), and the open-source ecosystem BL246 (Raspberry Pi CM5, expandable Hailo)—covering all high, mid, and low-end requirements.
Q3: Which industrial OS can EdgePLC run?
A: It has been adapted for CODESYS, Phoenix Contact vPLCnext, and WasomCodex, and also supports OpenPLC, Ubuntu, and Docker. It can run any Linux-compatible industrial software.
Q4: Is EdgePLC suitable for deploying industrial AI agents?
A: Yes. The built-in NPU supports local inference of vision models such as YOLO, CODESYS ensures control determinism on isolated cores, and AI inference results can directly drive I/O—the entire chain closes within one device.
Conclusion
EdgePLC is not “a PLC plus an AI box.” It is an open industrial hardware foundation where control, computing, and intelligence share the same system. For industrial OS vendors, AI agent developers, and control engineers, it removes the proprietary wall—and puts the choice of technology stack back in the user’s hands.
