VIGOR: Verifiable Industrial Geometry Optimization and Repair in Scene Programs

Key points

• Large language models (LLMs) increasingly generate structured layouts and executable scene programs from natural-language instructions.1
• However, executability does not imply engineering feasibility: an industrial scene program may execute successfully while its final state still violates coupled boundary, collision, clearance, aisle, or structural constraints.1
• We identify this mismatch as the executability-engineering-feasibility gap, where locally plausible placements can remain globally infeasible because corrections to one device may introduce violations elsewhere.1
• These results show that, under controlled industrial layout conditions, reliable scene-program generation requires not only executable programs, but also explicit verification and coordinated resolution of coupled engineering constraints.1

Sources (1)

• [1]VIGOR: Verifiable Industrial Geometry Optimization and Repair in Scene Programs
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 9, 07:34 AM

  We further propose VIGOR, which combines supervised scene-program generation with execution-grounded verification and staged geometric refinement.

  Large language models (LLMs) increasingly generate structured layouts and executable scene programs from natural-language instructions.

Extractive summary: sentences quoted from the sources.