Humans-in-the-Loop towards a more effective AI in manufacturing
HLOOP: Humans-in-the-Loop towards a more effective AI in manufacturing
Start 01/2023     Duration 18 months

Learn more



The adoption of Artificial Intelligence (AI) for Process Optimisation (PO) in manufacturing can significantly contribute to make the Irish manufacturing sector more competitive and sustainable. However, its adoption has been slower than expected. Looking at the way AI is being used in PO, the issue is the need of tailoring AI-solutions for every machine tool and context (used material, shape of the manufactured object). With this approach, the industry will need “tens of thousands’’ of unique AI- models, which is clearly infeasible.


Instead of developing an AI-solution for each machine tool and context, HLOOP proposes to build a single AI-model exploiting the synergy between AI and human-workers. Assessing the quality of the manufacturing process is more effectively done by humans than AI, while AI is better at dealing with high-dimensional decision problems. We plan to use the feedback of machine operators to train on-the-fly (that is without using any pre-collected dataset) and in real-time an AI-model that can predict when the manufacturing process is good/bad, and then optimise it.


A technological solution cannot be successful if it does not demonstrate explicit advantages on a complex manufacturing process. Laser Metal Additive Manufacturing, due its complexity and greenness, will be the testing ground for HLOOP

Financial support

This project is supported by the EU Commission Recovery and Resilience Facility under the Science Foundation Ireland Future Digital Challenge Grant Number 22/NCF/FD/10827.



Alessio Benavoli


Associate Professor

School of Computer Science and Statistics, TCD

Expertise: Probabilistic Machine Learning, Optimisation, Automatic control, Decision Making under Uncertainity, AI-based process optimisation, Smart Manufacturing.

Personal web page


Rocco Lupoi


Associate Professor

Dept. Mechanical, Manufacturing & Biomedical Engineering, TCD

Expertise: Additive Manufacturing (3D Printing), Laser Processing, Metallurgy, Post Processing, Powder Bed Fusion, Selective Laser Melting (SLM), Surface Engineering

Personal web page


Giacomo Cappelli

Research Assistant

Dept. Mechanical, Manufacturing & Biomedical Engineering, TCD

Expertise: Additive Manufacturing (3D Printing), Selective Laser Melting, Cold Spray Technology, Material Science, Microstructural Analysis, SEM, Solid Particle Erosion, Design of Experiments, CAD.

Personal web page


Eleni Zavrakli

Postdoc Researcher

School of Computer Science and Statistics, TCD

Expertise: Reinforcement Learning, Deep Neural Networks, Machine Learning, Dynamic Programming, Optimisation, Smart Manufacturing

Personal web page

Former member: Bobby Gillham


Have you ever faced the disappointment of your manufactured product not turning out as desired? The "recipe" you are using may not always produce the desired result, leaving you wondering which parameters to change next in order to maximize the product quality. HLOOP aims to automatise and simplify process optimisation in manufacturing. It uses your qualitative (or quantitative) feedback about the quality of your product to quickly optimise your manufacturing parameters and process.

Let's consider for example the process of baking a cake (figure-left). What is process optimisation in this case? Firstly, we carefully select the parameters to optimise, like amount of butter, sugar, and oven temperature. Next, we may experiment with four different recipes, each resulting in a different cake. We then taste each cake and rank the recipes based on our preferences. Using this valuable human feedback, Hloop can iteratively improve the recipe until we achieve the perfect cake (figure-right). Remarkably, this approach can be applied to any manufacturing process to achieve optimal products/processes, such as for instance in metal injection moulding (figure-centre).



Would you like to learn more about our theoretical framework for learning from human feedback in HLoop?

  • Benavoli, A., Azzimonti, D. and Piga, D. "Learning Choice Functions with Gaussian Processes." Proceedings of the 39th Conference on Uncertainty in Artificial Intelligence, 2023. [link]
  • Benavoli, A., Azzimonti, D. and Piga, D. "Bayesian Optimization For Choice Data" Proceedings of the Genetic and Evolutionary Computation Conference Companion (GECCO '23 Companion), 2023 [link]


Lets get in touch. Send us a message:

Trinity College Dublin

Email: alessio.benavoli@tcd.ie, lupoir@tcd.ie