Vendor-Aware Industrial Agents: RAG-Enhanced LLMs for Secure On-Premise PLC Code Generation
Researchers developed a Retrieval Augmented Generation (RAG) coding assistant specifically for Mitsubishi Electric's Structured Text (ST) PLC code. This system uses directed knowledge retrieval and prompt engineering to generate vendor-compliant control logic, overcoming the challenge of low-data domains where traditional LLMs fail on proprietary industrial dialects. The solution integrates compiler validation for real-time error correction and supports both cloud and local deployment.