Matteo Biagiola

About

Matteo Biagiola is a researcher whose work sits at the intersection of software testing and artificial intelligence, with a particular focus on ensuring the reliability and quality of Deep Reinforcement Learning (DRL) systems. His research addresses a critical and timely challenge: as RL-based agents move beyond game-playing into high-stakes real-world domains such as autonomous vehicles and robotics, rigorous testing methodologies become essential for safety and dependability. Biagiola's most influential contribution is his development of surrogate model-based testing approaches for DRL agents, a novel framework that has attracted over 35 citations across related publications and represents a significant methodological advance in AI quality assurance. Building on this foundation, his more recent work introduced μPRL, a mutation testing pipeline specifically designed for reinforcement learning systems and grounded in real-world faults — pushing the boundaries of how adequately RL agents can be evaluated before deployment. Earlier in his career, Biagiola also demonstrated an aptitude for applied engineering, contributing to the development of a specialized force feedback joystick for offshore robotic operations. Across these diverse contributions, his research consistently bridges theoretical rigor with practical impact, making him a noteworthy voice in the growing field of AI testing and verification.

Research Focus

Key Achievements

Top Papers

1. 1
   Testing of Deep Reinforcement Learning Agents with Surrogate Models

   29 citations · 2023
2. 2
   Testing of Deep Reinforcement Learning Agents with Surrogate Models

   7 citations · 2023
3. 3
   $\mu \text{PRL}$: A Mutation Testing Pipeline for Deep Reinforcement Learning Based on Real Faults

   4 citations · 2025
4. 4
   Development and integration of a novel IP66 Force Feedback Joystick for offshore operations

   4 citations · 2014

Key Collaborators

Contact & Links