The laboratory has excellent facilities for experimental fluid dynamics (including access to three large-scale fluid tunnels), and it is active in advancing optical methods for flow and transport measurements. We also have an extensive experience in fluid dynamics simulations and in developing novel numerical models.

We leverage our facilities to integrate our expertise in experimental fluid mechanics and machine learning. Our goals are

(i) to advance optical methods (particle image -resp. tracking- velocimetry, PIV -resp. PTV; background oriented Schlieren, BOS) and devise deep learning algorithms (e.g., specifically designed convolutional neural network, CNN) for real-time data analysis and control;

(ii) to devise AI multiagent systems capable of performing autonomous experiments and monitoring campaigns.

Our ambition is to design reliable frameworks for autonomous experimentation, which learn from the environment with without human supervision and can be applied beyond fluid dynamics for autonomous scientific discovery.

List of relevant projects:

• Machine Learning for fluid velocimetry and visualization
• Simultaneous PIV–LIF measurements using RuPhen and a color camera
• Droplet dispersion in airborne pathogen transmission
• Experimental characterization of droplet emission by turbulent puffs
• Model Cascades for Stochastic Particle Simulations of Rarefied Polyatomic Gases
• Autonomous optical velocimetry with a reinforcement learning agent

• Development of a computational fluid dynamic framework in OpenFOAM to study multiphase flow in porous media