LLM-ADAM: AI Framework Detects Pre-Print Anomalies in Additive Manufacturing
A research paper on arXiv proposes LLM-ADAM, a generalizable large language model (LLM) framework for detecting anomalies in G-code files before 3D printing. Additive manufacturing, particularly fused filament fabrication (FFF), has become accessible to labs and classrooms, but users may lack expertise to identify harmful settings in slicer profiles or G-code edits that affect extrusion, cooling, or adhesion. The framework decomposes anomaly detection into three roles: Extractor-LLM maps G-code to a structured process-parameter schema; Reference-LLM converts manufacturer specifications into reference parameters; and Checker-LLM compares extracted parameters against references to flag anomalies. This pre-print screening aims to prevent material waste and machine damage from accidental or adversarial errors. The paper is published on arXiv with ID 2605.03328.
Key facts
- LLM-ADAM is a generalizable LLM framework for pre-print anomaly detection in additive manufacturing.
- The framework uses three roles: Extractor-LLM, Reference-LLM, and Checker-LLM.
- Fused filament fabrication (FFF) has extended AM to laboratories, classrooms, and small production environments.
- Syntactically valid slicer profiles can encode thermally or geometrically harmful settings.
- Subtle G-code edits can alter extrusion, cooling, or adhesion before a print begins.
- Pre-print G-code screening catches accidental or adversarial machine-program errors.
- The paper is available on arXiv with ID 2605.03328.
- The framework aims to save material and machine time.
Entities
Institutions
- arXiv