PLCWorld
Research benchmark of 100 Structured Text tasks graded in closed loop, with task success and safety reported separately.
Who it's for
People comparing published PLC-code or controller-agent results. The number on this page is only as good as the paper or site it cites.
Caveat / limitations
Paper: Yunji Kim, Yunseok Lee, Hyunwoo Seo, Jaerim Choi, and Woojin Lee, arXiv:2610.02982. 100 synthetic tasks and 473 task-condition pairs across motion control and material handling. Direct GPT-5.5 is reported at 82.70% task success on Easy cases and 25.10% on Hard cases. AutoFactory-Bench adapts this package and says its audited scores are not the same numbers. The project page is the code link given in the paper.
Protocols
Sources
- Project pagehttps://yunji0516.github.io/PLCWorld/
- arXivhttps://arxiv.org/abs/2610.02982
This page: https://otaihub.com/tools/plcworld
Same category
- AutoFactory-BenchClosed-loop benchmark: agents write one Structured Text file for 100 factory tasks and a simulator grades success and safety.
- Agents4PLCMulti-agent research framework and a verifiable Structured Text benchmark with formal specs and reference code.
- LLM4PLCResearch pipeline that loops an LLM with grammar checks, a compiler, and an SMV verifier to produce Structured Text.
- AutoPLCResearch generator for vendor-specific Structured Text, evaluated on Siemens SCL and CODESYS with a compiler in the loop.