Haolin (Oliver) Fan

Haolin (Oliver) Fan

Creating digital reflections that make physical systems smarter.

Publications

You can also find my articles on my Google Scholar profile.

Journal Articles

Sim2Know: new paradigm of digital twins to design and inform human-centric knowledge system

Published in CIRP Annals, 2025

Sim2Know is a framework that addresses limited real-world data and implicit knowledge capture by using a digital twin to generate synthetic data and a hybrid training method combining transfer learning and data augmentation. It achieves 90.31% precision in recognizing key human actions in metal additive manufacturing and enhances knowledge systems by contextualizing human-machine interactions.

Recommended citation: Li, Bingbing, et al. "Sim2Know: new paradigm of digital twins to design and inform human-centric knowledge system." CIRP Annals (2025).
Download Paper

MetalMind: A Knowledge Graph-Driven Human-Centric Knowledge System for Metal Additive Manufacturing

Published in npj Advanced Manufacturing, 2025

This paper introduces a human-centric knowledge system for Industry 5.0 that integrates expert and formal knowledge to support decision-making and workforce development. Key innovations include: (1) an LLM-based knowledge graph pipeline with collaborative verification, (2) a hybrid retrieval framework outperforming traditional methods, and (3) an MR interface for real-time interaction. A metal additive manufacturing case study demonstrates its effectiveness in enhancing expertise, retrieval, and human-machine collaboration.

Recommended citation: Fan, Haolin, et al. "MetalMind: A Knowledge Graph-Driven Human-Centric Knowledge System for Metal Additive Manufacturing." (2025).
Download Paper

MaViLa: Unlocking new potentials in smart manufacturing through vision language models

Published in Journal of Manufacturing Systems, 2025

The paper introduces MaViLa, a vision-language model tailored for smart manufacturing, which improves visual understanding through a retrieval-augmented dataset creation and a two-stage training process. MaViLa outperforms general-purpose models in tasks like process optimization and quality control, supporting better decision-making and integration with external tools, thus advancing autonomous manufacturing systems.

Recommended citation: Fan, Haolin, et al. "MaViLa: Unlocking new potentials in smart manufacturing through vision language models." Journal of Manufacturing Systems 80 (2025): 258-271.
Download Paper

AutoMEX: Streamlining material extrusion with AI agents powered by large language models and knowledge graphs

Published in Materials & Design, 2025

AutoMEX is a framework that integrates large language models and a knowledge graph to automate the material extrusion (MEX) additive manufacturing process. It enhances workflow integration, from CAD design to machine operation, with minimal human input. Experiments show improved print strength and high user acceptance.

Recommended citation: Fan, Haolin, et al. "AutoMEX: Streamlining Material Extrusion with AI Agents Powered by Large Language Models and Knowledge Graphs." Materials & Design (2025): 113644.
Download Paper

New era towards autonomous additive manufacturing: a review of recent trends and future perspectives

Published in International Journal of Extreme Manufacturing, 2025

This paper reviews current intelligent additive manufacturing (IAM) systems, highlighting limitations like fragmented AI use and poor human-machine interaction. It proposes a shift to autonomous AM via a four-layer hierarchical framework that enables machines to independently analyze and execute tasks.

Recommended citation: Fan, Haolin, et al. "New era towards autonomous additive manufacturing: a review of recent trends and future perspectives." International Journal of Extreme Manufacturing (2025).
Download Paper

Enhancing metal additive manufacturing training with the advanced vision language model: A pathway to immersive augmented reality training for non-experts

Published in Journal of Manufacturing Systems, 2024

This paper presents a novel training system for the Renishaw AM400 metal printer that combines Vision Language Models (VLM), Augmented Reality (AR), and Digital Twins (DT). The system enhances recognition, user interaction, and accessibility for non-experts, offering a new benchmark for smart manufacturing training.

Recommended citation: Fan, Haolin, et al. "Enhancing metal additive manufacturing training with the advanced vision language model: A pathway to immersive augmented reality training for non-experts." Journal of Manufacturing Systems 75 (2024): 257-269.
Download Paper

Embodied intelligence in manufacturing: leveraging large language models for autonomous industrial robotics

Published in Journal of Intelligent Manufacturing, 2024

This paper explores using large language model (LLM) agents in industrial robotics, focusing on autonomous design, decision-making, and task execution. It introduces a framework with three key components: task-parameter matching, autonomous tool path design, and integration with robotic simulations. Results show GPT-4 excels in task planning, achieving an 81.88% success rate in task completion.

Recommended citation: Fan, Haolin, et al. "Embodied intelligence in manufacturing: leveraging large language models for autonomous industrial robotics." Journal of Intelligent Manufacturing 36.2 (2025): 1141-1157.
Download Paper

Conference Papers

Unleashing the Potential of Large Language Models for Knowledge Augmentation: A Practical Experiment on Incremental Sheet Forming

Published in Procedia Computer Science, 2024

This research is presented at the International Conference on Industry 4.0 and Smart Manufacturing (ISM) 2024. It explores how Large Language Models (LLMs) can be adapted to better understand and operate within the Incremental Sheet Forming (ISF) domain. By developing an automated pipeline for extracting and enriching ISF-specific knowledge, and fine-tuning models like Alpaca-33B, the study achieves a 10.4% improvement in domain knowledge acquisition over GPT-3.5. A new conversational prototype further boosts accuracy and relevance, paving the way for advanced ISF applications such as knowledge graphs and quality prediction.

Recommended citation: Fan, Haolin, et al. "Unleashing the potential of large language models for knowledge augmentation: A practical experiment on incremental sheet forming." Procedia Computer Science 232 (2024): 1269-1278.
Download Paper

CAtNIPP: Context-aware attention-based network for informative path planning

Published in Proceedings of Machine Learning Research, 2024

This work is orally presented at the 6th Conference on Robot Learning (CoRL) 2024. Informative Path Planning (IPP) is a challenging problem that requires balancing exploration and exploitation under resource constraints. CAtNIPP introduces a fully reactive, deep reinforcement learning-based solution using self-attention to guide efficient path planning. By learning to form a global context and make adaptive local decisions, CAtNIPP outperforms traditional IPP methods in both performance and computation speed, with demonstrated results on real hardware.

Recommended citation: Cao, Y., Wang, Y., Vashisth, A., Fan, H. & Sartoretti, G.A.. (2023). CAtNIPP: Context-Aware Attention-based Network for Informative Path Planning. Proceedings of The 6th Conference on Robot Learning, in Proceedings of Machine Learning Research 205:1928-1937
Download Paper