Sample-Based Predictive Control for Agile Robots: Fast planning, robust execution, and safe operation

Relatore
 Prof. Valerio Modugno - University College London, UK

Data
 12-mag-2026 - Ora: 17:00 Sala Verde (presenza ed on line)

Abstract: This seminar offers an implementation-minded view of stochastic, sample-based predictive control designed for agile robotics. It explores how Model Predictive Path Integral (MPPI) control efficiently handles nonconvex objectives and discontinuous events without relying on gradients. By utilizing parallel rollouts on GPUs, MPPI ensures explicit, predictable latency budgets. The talk will cover recent architectural advancements, including GPU-accelerated MPPI for 12-DoF quadrupeds, Feedback-MPPI utilizing rollout differentiation for high-rate control, and BC-MPPI for enforcing probabilistic constraints safely without heavy penalties.
 
Short Bio: Valerio Modugno has been a Lecturer in the Robotics and AI groups at UCL's Department of Computer Science since 2024. His research interests comprise Humanoid Whole-Body Control, Optimal Control, teleoperation for legged robots, Reinforcement Learning, Black-Box Optimization, and safety for control and learning strategies. Before joining UCL in 2022, he was a Post-Doctoral Researcher at Sapienza University of Rome, where he also received his Ph.D. in 2017. During his studies, he was a visiting researcher at the Technical University of Darmstadt in 2014 and at INRIA Grand-Est Nancy in 2015.
 
Modalitá ibrida, link zoom: https://univr.zoom.us/j/85630652158
Referente: Dr. Daniele Meli
 
Data pubblicazione
30-apr-2026

Referente
Daniele Meli
Dipartimento
Informatica