Marcello Restelli
Papers
24
Total Citations
372
H-Index
10
About
Marcello Restelli is a leading researcher in robotics and reinforcement learning, whose work bridges the gap between reliable robot perception and advanced decision-making algorithms. His early contributions focused on dead-reckoning localization for mobile robots, where he pioneered the use of optical mice as kinematic-independent odometric sensors—a novel approach that eliminates slipping errors and provides robust indoor navigation. This foundational work, including his most-cited paper (83 citations), has influenced practical robotics design. Restelli’s major impact, however, lies in multi-objective reinforcement learning (MORL). He introduced methods for approximating continuous Pareto manifolds (51 citations), enabling agents to optimize multiple conflicting objectives simultaneously—critical for real-world control in robotics and economics. He also advanced policy gradient methods with adaptive step-size techniques (48 citations) and safe policy gradients (20 citations), improving stability and safety in trial-and-error learning. His work on batch RL for controlling a mobile wheeled pendulum robot (15 citations) demonstrates direct robotic application. With over 300 total citations across these seminal papers, Restelli’s research continues to shape how autonomous systems learn and navigate complex environments.
Research Focus
Key Achievements
Top Papers
- 1
- 2A kinematic-independent dead-reckoning sensor for indoor mobile robotics52 citations · 2005
- 3
- 4Adaptive Step-Size for Policy Gradient Methods48 citations · 2013
- 5Smoothing policies and safe policy gradients20 citations · 2022
- 6Batch Reinforcement Learning for Controlling a Mobile Wheeled Pendulum Robot15 citations · 2008
- 7DEAD RECKONING FOR MOBILE ROBOTS USING TWO OPTICAL MICE14 citations · 2004
- 8MRT: Robotics Off-the-Shelf with the Modular Robotic Toolkit10 citations · 2007
- 9Concepts for Anchoring in Robotics10 citations · 2001
- 10