Marco Pardini
PhD Student in Smart Industry @ UniPi
My research bridges the gap between Artificial Intelligence and Robotics, focusing on making humanoid robots hyper-realistic in real-time, with a specific emphasis on facial control. I leverage Unreal Engine for advanced simulations and biosignals to decode human emotional states. My ultimate goal is to embed this emotional awareness into cognitive architectures, enabling humans to interact seamlessly with robots that genuinely understand and respond to the emotional state of their interlocutors.
Research Groups & Affiliations
MLPI Group
Machine Learning and Process Intelligence Group, Pisa
Centro Piaggio
Affiliated Researcher
BRAVE Project
Project Member
My PhD research sits at the interdisciplinary crossroads of Computer Engineering and Biomedical Engineering, jointly supervised by Prof. Mario Cimino and Prof. Alberto Greco. Prof. Cimino leads research in deep learning, explainable AI, computer vision, and cognitive architectures. Conversely, Prof. Greco is a leading expert in electrodermal signal analysis and affective computing. Operating precisely at the intersection of these two domains, my specific focus is applying advanced deep learning and computer vision algorithms directly to robotic platforms—bridging the gap between cognitive artificial intelligence and physiological affective response.
Cognitive Humanoid Robotics
Working directly with advanced humanoid platforms, focusing on hyper-realistic control, collision avoidance, and facial retargeting in real-time to create believable interactions.
Bringing Humanoids to Life
Abel is a new generation hyper-realistic humanoid robot, conceived as a research platform for social interaction, emotion modeling, and embodied intelligence. Resembling an 11–12 year old boy, Abel is a unique collaboration between the University of Pisa and Gustav Hoegen (Biomimic Studio, London). Physically, Abel comprises a head and upper torso driven by 42 high-end servo motors (Futaba, MKS, Dynamixel). The head alone houses 21 motors dedicated to facial expressions, gaze, and speech simulation, allowing Abel to express a wide spectrum of human emotions. The robot is also equipped with an integrated torso camera, binaural microphones to emulate human acoustic perception, and an internal speaker.
Seth, born from the same collaboration, relies exclusively on an array of 13 Dynamixel servo motors. These grant extremely precise, sub-millimeter control over its intricate facial movements.
My primary objective within this domain is absolute control over these complex mechatronic systems. I began by entirely rewriting Abel's control and movement architecture from scratch. Now, leveraging Seth's extreme mechanical precision, I am utilizing Deep Learning to perform real-time, high-fidelity facial retargeting, pushing the boundaries of what is possible in human-robot affective interaction.
Interdisciplinary Research & Simulation
Complementing my work in humanoid robotics, my research extends into affective computing and high-fidelity virtual environments, driving the development of a comprehensive ecosystem for human-machine interaction.
Affective Computing
In our initial work on continuous assessment of Social Anxiety Disorder (SAD) in Virtual Reality, we focused on multimodal classification. By integrating continuous self-ratings with objective physiological data (ECG/EDA), our Late Fusion Transformer framework successfully classified high and low anxiety groups, achieving an F1-score of 0.853 and 82.5% accuracy.
We are now tackling the significantly more complex challenge of multivariate continuous regression to map dynamic, in-the-moment physiological signals directly to the user's fluctuating emotional state.
Unreal Engine & MetaHumans
The process of reviving historical figures begins with high-density point clouds, meticulously converted into workable static meshes directly within Unreal Engine. Advanced grooming tools recreate fine details, followed by rigging intricate models to the MetaHuman framework.
These digital avatars can be projected into physical Holoboxes. Displaying MetaHumans holographically requires a complex multi-camera setup within Unreal Engine to render specific angles simultaneously, constructing a seamless 3D illusion.
Education
PhD in Smart Industry
Nov 2024 - PresentUniversity of Pisa
M.Sc. Artificial Intelligence & Data Engineering
2022 - 2024University of Pisa | 110/110 cum laude
Thesis: Service-Oriented Cognitive System for Social Robotics orchestrated by LLMs
B.Sc. Computer Engineering
2019 - 2022University of Pisa | 110/110 cum laude
Publications
From Google Scholar
Featured Award
IEEE MetroXraine 2025
Best Graphical Abstract