M.Sc. candidate in Artificial Intelligence and Robotics, Sapienza University of Rome. B.Sc. in Electronic Engineering.
I work on dynamical systems like manipulators, drones and multi-agent networks. I also work on neural networks applied to physical measurements.
Most of my repositories are implementations of published methods and specific algorithms. I write them from the equations in the paper and then test them against a baseline or a known solution, to see whether they actually work. Three to start with:
- aerial-robotics-project — decoupled dynamics of an unmanned aerial manipulator: tube-based LPV-MPC for a hexarotor carrying a 3-link arm, from Eskandarpour et al., IEEE TAES 2025, compared with the ERTF baseline.
- median-consensus-multi-agent-system — non-smooth median consensus tracking for open networks, from the IEEE TAC paper: Filippov solutions, Clarke gradients, finite-time bounds. It runs on synthetic references or on the six PM2.5 sensors from CNN-and-RNN-regression, so those bounds can be checked against real noisy data.
- lhc-event-classification — six hand-built physics features against the raw detector images on the same events: a CNN on the images reaches 0.8712, a random forest on the features 0.8562.
The rest are pinned below, or in the repository list.
Rome, Italy · giorgio0420.github.io · desantisgiorgio20@gmail.com