LLM-ADAM: A Generalizable LLM Agent Framework for Pre-Print Anomaly Detection in Additive Manufacturing
Additive manufacturing (AM) is revolutionizing the landscape of modern manufacturing, providing the ability to produce complex geometries on demand across various industries. The advent of fused filament fabrication (FFF) technology has made AM accessible to laboratories, classrooms, and small-scale production environments. However, this accessibility places the responsibility of process planning in the hands of users who may lack specialized manufacturing knowledge. The challenge lies in the fact that even a syntactically valid slicer profile can include thermally or geometrically detrimental settings, and minor modifications in G-code can significantly impact extrusion, cooling, or adhesion prior to the printing process.
To combat these challenges, pre-print G-code screening emerges as a crucial step in identifying accidental or adversarial machine-program errors before any material or machine time is wasted. In this context, the paper introduces LLM-ADAM, a generalizable large language model (LLM) framework specifically designed for pre-print anomaly detection in additive manufacturing.
Framework Overview
The LLM-ADAM framework decomposes the anomaly detection task into three distinct roles:
- Extractor-LLM: This component maps a G-code file to a structured process-parameter schema, enabling a systematic understanding of the parameters involved in the printing process.
- Reference-LLM: This role converts printer and material documentation into aligned operating ranges, ensuring that the printing settings adhere to the specifications of the materials and printers being utilized.
- Judge-LLM: This segment interprets a deterministic deviation table alongside G-code evidence to ascertain whether a part is non-defective or falls into an anomaly class.
Importantly, the framework allows for interchangeability among printers, materials, and LLM backbones, treating these elements as test conditions rather than fixed assumptions.
Evaluation and Results
The effectiveness of the LLM-ADAM framework was evaluated using a corpus of 200 FFF G-code samples, which encompassed two different desktop printer families, two materials, and five defect classes including non-defective, under-extrusion, over-extrusion, warping, and stringing. The results revealed that the best configuration of the framework achieved an impressive accuracy rate of 87.5%. In contrast, the strongest engineered single-LLM baseline achieved only 59.5% accuracy.
The findings suggest that structured decomposition of tasks is a more significant contributor to performance improvement than the strength of the LLM backbone alone. Notably, the framework was able to identify defect classes at or near the ceiling for leading configurations. However, residual errors were primarily concentrated in conservative false alarms for non-defective samples.
Conclusion
The introduction of LLM-ADAM represents a significant advancement in the field of additive manufacturing, providing a robust framework for pre-print anomaly detection. By leveraging structured decomposition and the versatility of LLMs, this framework enhances the reliability and efficiency of the AM process, ensuring that users can produce high-quality prints without the risk of costly errors. As additive manufacturing continues to grow, frameworks like LLM-ADAM will be essential in bridging the gap between accessibility and expertise in this rapidly evolving domain.
Related AI Insights
- Spectral Structure & Equivalence in Multilabel Fisher Discriminants
- Whoop vs Fitbit Air: Best Fitness Band Compared 2024
- AI Chatbot Eases Breakup Pain: Study Shows Lasting Relief
- MAGE: Protecting LLM Agents from Long-Horizon Threats
- Copula Correction for Robust Treatment Effect Estimation
- Cryptographic Defense Against Dependency Confusion Attacks
- Confidential Computing for Secure Agentic AI Systems
- DGPO: Advanced Policy Optimization for Precise Credit Assignment
- 2025 LLM Hackathon: Advances in Materials Science & Chemistry
- Lenovo Pro 9i Aura vs Dell XPS: Best Premium Laptop 2024
