Papers
20
Total Citations
265
H-Index
10
About
Enrico Marchesini is a robotics and artificial intelligence researcher whose work sits at the intersection of deep reinforcement learning (DRL), safe autonomy, and multi-robot systems. He has made significant contributions to mapless navigation, pioneering approaches that demonstrate discrete action-space algorithms—such as Double Deep Q-Networks with parallel asynchronous training—can rival continuous alternatives, a finding that has garnered over 70 citations. His research consistently pushes the boundaries of safe reinforcement learning, introducing formal verification methods to guarantee neural network behavior in safety-critical contexts, including autonomous robotic-assisted surgery, where his work has attracted nearly 40 citations. Marchesini has also advanced multi-robot coordination by developing centralized training frameworks and evolutionary policy search strategies that enhance cooperation without sacrificing exploration. His proposal of a novel aquatic navigation benchmark further reflects his commitment to rigorous, real-world evaluation of safe DRL systems. Across his portfolio, he integrates evolutionary algorithms with gradient-based learning—through techniques like Genetic Soft Updates—to improve policy robustness. With over 200 cumulative citations across a focused body of work, Marchesini represents an emerging voice shaping how autonomous systems can learn reliably and safely in complex, dynamic environments.
Research Focus
Key Achievements
Top Papers
- 1Discrete Deep Reinforcement Learning for Mapless Navigation73 citations · 2020
- 2
- 3
- 4
- 5Benchmarking Safe Deep Reinforcement Learning in Aquatic Navigation14 citations · 2021
- 6
- 7Curriculum learning for safe mapless navigation13 citations · 2022
- 8
- 9Genetic Soft Updates for Policy Evolution in Deep Reinforcement Learning12 citations · 2021
- 10Genetic Deep Reinforcement Learning for Mapless Navigation11 citations · 2020