Andrea Protopapa

Ph.D. candidate in Intelligenza Artificiale , 39^th cycle (2023-2026)
Department of Control and Computer Engineering (DAUIN)

andrea.protopapa@polito.it

Profile

PhD

Research topic

My research interests are focused on learning for compliant robots, developing data-efficient Sim-to-Real transfer for soft robots and IRL methods (image-based and graph-based) for long-horizon tasks.

Tutors

Research presentation

Poster

Research interests

Robotics and human-robot interaction for Industry 4.0

Knowledge representation and reasoning

Biography

I am a PhD student in the National PhD in Artificial Intelligence for Industry at Politecnico di Torino (XXXIXthCycle), working in the VANDAL Lab under the supervision of Professor Giuseppe Averta. My research focuses on learning skills for compliant robots operating in complex, unstructured environments, with two main lines: Sim-to-Real transfer and long-horizon task learning from demonstrations.
On the Sim-to-Real side, I am the lead author of RF-DROPO, a reset-free offline Adaptive Domain Randomization method tailored to soft robot control. RF-DROPO infers a distribution over dynamics parameters from a small amount of real data and uses it for informed domain randomization in simulation, enabling zero-shot transfer of contact-rich manipulation policies to a physical soft trunk robot and outperforming existing DR baselines. This work builds on my earlier contributions on domain randomization for soft robots, including a paper at IROS 2023.
In parallel, I work on Inverse Reinforcement Learning (IRL) for long-horizon, multi-object manipulation tasks. I have explored image-based IRL with progress embeddings, subtask decomposition, and intrinsic rewards, and more recently proposed a graph-based IRL framework that uses object-centric scene graphs and Graph Neural Networks to automatically discover subtasks and learn step-shaped procedural rewards.
I also enjoy teaching and mentoring, having contributed to courses in Robot Learning, Machine Learning and Deep Learning, and Algorithms and Data Structures, and supervised an MSc thesis. I have been a visiting researcher at INRIA Lille and received the SOFA Organizers’ Prize and a Best Poster Award at the SmartData@PoliTO Workshop for my work on robust learning for soft robots. I am planning an Internship at TU Darmstadt, with co-supervision of Professor Georgia Chalvatzaki.

Awards and Honors

Organizers’ Prize SOFA Award, 2023/11/14, International SOFA Symposium 2023, Paris, France https://www.sofa-framework.org/applications/plugins/reinforcement-learning-with-domain-randomization/ (2023)
1st Year PhD Best Poster Award, 2023/09/25, 9th SmartData@PoliTO Workshop, Loano, Italy (2023)

Teaching

Teachings

Master of Science

Robot Learning. A.A. 2025/26, INGEGNERIA INFORMATICA (COMPUTER ENGINEERING). Collaboratore del corso
Machine learning and Deep learning. A.A. 2024/25, DATA SCIENCE AND ENGINEERING. Collaboratore del corso
Robot Learning. A.A. 2024/25, INGEGNERIA INFORMATICA (COMPUTER ENGINEERING). Collaboratore del corso
Robot Learning. A.A. 2023/24, INGEGNERIA INFORMATICA (COMPUTER ENGINEERING). Collaboratore del corso
Machine learning and Deep learning. A.A. 2023/24, DATA SCIENCE AND ENGINEERING. Collaboratore del corso

MostraNascondi A.A. passati

Bachelor of Science

Algoritmi e strutture dati. A.A. 2024/25, INGEGNERIA INFORMATICA. Collaboratore del corso
Algoritmi e strutture dati. A.A. 2023/24, INGEGNERIA INFORMATICA. Collaboratore del corso