My long-term goal is to create physical and virtual agents that can learn to live, interact and communicate with humans in dynamic environments/domains. These agents should help people to live an overall better life. For this reason, my research is focused on reinforcement learning, AI explainability, robot control and knowledge representation.


I am the founder and coordinator of the KRL Group, the Knowledge, Reasoning and Learning Research Group in the Cognitive Cooperating Robots Laboratory, at Sapienza University of Rome. All of my current research projects involve this group. Similarly, all current and past students are (or were) a part of the KRLGroup. A list of publications from this team is available here.


Ongoing Research

  • Reinforcement Learning 

A large part of the my (and my team’s) research both in academia and industry is focused on reinforcement learning (RL). I am specifically interested in sparse-reward settings, model-based RL and imitation learning. We apply fundamental research on this topic to several domains, with a specific focus to robotics.

  • Explainable AI

One of the main interests at KRL Group is Explainable AI (XAI). This is an important area of research with several applications in a multitude of fields and domains. We focus, for example, on explainability in deep (and reinforcement) learning applied to computer vision, natural language understanding, chemistry, and more to come.

  • Robot Learning

We appl

  • Space Robotics & AI


Past Research

  • COVID-19 

The year 2020 saw the COVID-19 virus lead to one of the worst global pandemics in history. As a result, governments around the world have been faced with the challenge of protecting public health while keeping the economy running to the greatest extent possible. The main focus of this research consisted in developing a novel agent-based pandemic simulator, as well as a reinforcement-learning based methodology for optimizing fine-grained mitigation policies within this simulator and a Hidden Markov Model for predicting infected individuals based on partial observations regarding test results, presence of symptoms, and past physical contacts. More details are available in these workshop, conference and journal papers.

  • Semantic Mapping