Michael Vanuzzo
Postdoctoral Research Fellow | Machine Learning & Collaborative Robotics
I am a Postdoctoral Research Fellow at the University of Padova (Dept. of Management and Engineering - DTG), specializing in Machine Learning, Physics-Informed Deep Learning, and Collaborative Robotics.
My research bridges artificial intelligence and physical systems: from developing physics-informed temporal deep learning architectures for Remaining Useful Life (RUL) prediction and battery State of Health (SOH) estimation in the Horizon Europe ECS4DRES project, to pioneering real-time, context-aware 3D Human Motion Prediction (HMP) for safe, anticipatory Human-Robot Collaboration (HRC).
Core Research Areas
Connecting computational machine learning with physical robotics, real-time sensing, and industrial domain knowledge.
Physics-Informed Deep Learning
Hybrid physics-data neural architectures (BiLSTM, Causal TCNs) integrating electro-thermal degradation models (Coffin-Manson, Palmgren-Miner) for Remaining Useful Life (RUL) prediction of IGBT power modules and battery State of Health (SOH) estimation.
Real-Time Human Motion Prediction
Pioneering context-aware 3D deep learning models (RNNs, GNNs, LLMs) to shift Human-Robot Collaboration from reactive safety stops to proactive anticipatory interaction (Observe-Predict-Plan-Act).
Digital Twins & Cobot Systems
Experimental cobot validation, cross-platform XR visual feedback with 3D human avatars (HoloLens, Magic Leap 2), and plant-wide discrete-event simulation digital twins.
Work Experience
Academic research positions, international visiting stays, industrial engineering contracts, and university teaching.
Postdoctoral Research Fellow
- Conducting research on Machine Learning applied to power electronics and energy storage within the Horizon Europe ECS4DRES project (Grant No. 101139790).
- Developing Physics-Informed Deep Learning architectures (BiLSTM, Causal TCNs) for Remaining Useful Life (RUL) prediction of IGBT power modules, achieving zero-shot model generalization across variable multi-stage thermal stress mission profiles.
- Integrated physical priors and empirical degradation models (Coffin-Manson law, Palmgren-Miner damage rule, electro-thermal normalization) into end-to-end deep learning pipelines.
- Engineered containerized MLOps development environments (Docker, Apptainer/Singularity, PyTorch, WandB) and automated telemetry pipelines to accelerate parallelized multi-GPU training on High-Performance Computing (HPC) clusters.
- Providing technical consulting and scientific supervision for a research project on State of Health (SOH) estimation of VRLA batteries in Uninterruptible Power Supplies (UPS) using Electrochemical Impedance Spectroscopy (EIS) and CNNs.
Ph.D. Researcher in Mechatronics (AI & Robotics)
Visiting Ph.D. Researcher
Research Software Engineer — MICS Project
Computer Vision & AR R&D Consultant
Teaching Assistant & Academic Tutor
Academic Thesis Supervisor & Co-Advisor
Process Simulation Engineer
Control Systems Engineer
Mechanical Engineering Trainee & Drafter
Publications & Software
International peer-reviewed journals (IEEE TAI, IEEE RA-L), conference proceedings (IFAC, IEEE CASE, ERF), and open-source robotics frameworks.
Transfer Learning for Human Motion Prediction: Improving Accuracy under Data Scarcity
Human Motion Prediction Using Spatial Semantics of Objects: The PADSO Approach
Enhancing Real-Time Body Pose Estimation in Occluded Environments Through Multimodal Musculoskeletal Modeling
Enhancing Robot Collaboration by Improving Human Motion Prediction Through Fine-Tuning
Human Motion Prediction Metrics: From Time to Frequency
Towards Explainable Human Motion Prediction in Collaborative Robotics
RT-HMP: A Modular and Extensible Framework for Real-Time Human Motion Prediction
Education & Recognition
Doctoral training, international visiting stays, competitive European grants, and academic accolades.
Degrees & Formation
Ph.D. in Mechatronic Engineering
Researched 3D context-aware Deep Learning architectures (RNNs, GNNs, LLMs) and designed RT-HMP, an open-source ROS 2 framework for real-time human motion prediction enabling safe, proactive human-robot collaboration.
Master of Science (MSc) in Mechatronic Engineering
Specialized in advanced control systems, industrial automation, robotics, and software engineering for high-reliability applications.
Bachelor of Science (BSc) in Mechanical & Mechatronic Engineering
Core curriculum in mechanical design, structural analysis, electronics, dynamical systems, and mathematical modeling.
High School Diploma in Mechanics, Mechatronics & Energy
Specialization in Mechanics and Mechatronics curriculum.
Honors, Grants & Awards
Advanced Training & Summer Schools
Robotics, Control Systems & AI Summer School
Advanced topics in Computer Vision, Human-Robot Interaction, autonomous navigation, and assistive robotics manipulation.
ACM SIGSOFT Summer School for Software Engineering in Robotics
Robotics software engineering, robot swarm design, formal system validation techniques, and practical algorithm implementation.
DeepLearn 2023 Summer School
Comprehensive sessions on deep neural architectures, generative models, and state-of-the-art AI applications.
Technical Skills & Tooling
Core toolchains, software stacks, frameworks, and engineering methodologies utilized across research projects and industry collaborations.
AI, Machine Learning & Deep Learning
Robotics & Automation
XR, Simulation & Digital Twins
Software Engineering & MLOps
Languages
Professional Memberships
- ✓ IEEE (Institute of Electrical and Electronics Engineers), Member No. 98708094
- ✓ IEEE Robotics & Automation Society (RAS), Member
- ✓ IEEE Young Professionals, Member