News_2025

 December

• I’m thrilled to share that I’ve been awarded a Fondo Italiano per la Scienza (FIS 3) Starting Grant funded by the Italian Ministry of Universities and Research (MUR) for my 5-year project, “Real-World Autonomous Agents through Trustworthy Reinforcement Learning.” This funding will allow me to kickstart my research lab, focusing on developing cooperative and trustworthy RL agents for real-world deployment. The project aims to lay down the foundations of RL methods for complex and realistic settings, including robotic warehouses and power grids ⚙️⚡, where agents must operate under partial information, collaborate with humans and other agents 🤝, and provide formal safety guarantees 🛡️

 November

• “Verifying Online Safety Properties for Safe Deep Reinforcement Learning” is finally published in ACM TIST and starts to investigate how to combine verification approaches for improving the safety of RL agents.

 September

• Our work “Designing Control Barrier Function via Probabilistic Enumeration for Safe Reinforcement Learning Navigation” has been accepted for publication in the IEEE Robotics and Automation Letters.

 January

• “Improving Policy Optimization via ε-Retrain” investigates how to improve an agent’s safety and performance by retraining from unsafe states visited at training time. Check it out at AAMAS 2025.
• Our latest work fixing multi-agent value factorization formalization, “On Stateful Value Factorization in Multi-Agent Reinforcement Learning” has also been accepted at AAMAS 2025.
• A new collaboration with Alp Aydeniz, Robert Loftin, Christopher Amato, and Kagan Tumer has been accepted at AAMAS 2025. Check out our abstract “Safe Entropic Agents under Team Constraints”!