Correct-by-Construction G-Code Generation: A Neuro-Symbolic Approach via Separation Logic

2026-05-11

AI Summary

This work addresses safety hazards in G-code generation caused by physical collisions and the lack of formal verification by proposing a self-correcting generation framework that integrates a neural language model (GLLM) with a separation logic verifier. For the first time, separation logic is applied to neural code generation, modeling physical space via a Spatial Heap and interpreting collisions as spatial resource conflicts. When verification fails, the framework provides precise feedback based on minimum bounding boxes to guide the GLLM in iteratively refining its output. This approach enables the automatic synthesis of provably collision-free G-code, substantially reducing manual intervention and enhancing the safety and reliability of autonomous manufacturing systems.

Abstract

This paper proposes a neuro-symbolic framework for G-code generation by integrating the GLLM neural method (Abdelaal et al., 2025) with our established Separation Logic (SL) verifier. We introduce a two-component architecture where GLLM serves as a creative generator and the SL Prover, utilizing the Spatial Heap model, acts as a deterministic verifier. By defining physical collisions as logical Spatial Data Races - violations of the separating conjunction in SL - the framework translates proof failures into structured mathematical feedback. These failures are condensed into minimal bounding boxes that act as precise spatial directives for GLLM's iterative self-correction. This synergy establishes a self-correcting generative cycle that reduces the need for manual oversight, supporting the production of verified G-code to enhance safety in autonomous manufacturing.

Problem

• G-code generation
• correctness
• collision avoidance
• autonomous manufacturing
• safety

Innovation

• Neuro-Symbolic
• Separation Logic
• G-Code Generation
• Spatial Data Race
• Self-Correcting Generation