Paolo Tonella
Papers
4
Total Citations
42
H-Index
3
About
Paolo Tonella is a distinguished software engineering researcher whose work sits at the cutting edge of artificial intelligence testing and verification. His research focuses on ensuring the reliability and safety of deep reinforcement learning (DRL) systems, with particular emphasis on autonomous vehicles, robotics, and other safety-critical applications where the stakes of software failure are exceptionally high. Tonella's most significant contributions center on developing rigorous testing methodologies for DRL agents — a notoriously difficult challenge given the complexity and non-deterministic nature of learned behaviors. His highly cited work on surrogate model-based testing of DRL agents (accumulating over 36 citations) introduced innovative frameworks that evaluate agent quality without requiring exhaustive real-world interaction. Building on this foundation, his recent μPRL pipeline applies mutation testing principles derived from real-world faults, bringing traditional software engineering rigor into the machine learning domain. Particularly noteworthy is Tonella's commitment to bridging academic research and industrial practice, exemplified by the real-world deployment of his Surrealist framework for testing industrial robot navigation systems. This translational impact demonstrates that his contributions extend well beyond theoretical advances, directly shaping how industry validates intelligent autonomous systems in dynamic, unpredictable environments.
Research Focus
Key Achievements
Top Papers
- 1Testing of Deep Reinforcement Learning Agents with Surrogate Models29 citations · 2023
- 2Testing of Deep Reinforcement Learning Agents with Surrogate Models7 citations · 2023
- 3
- 4