Lorenzo Rosasco

About

Lorenzo Rosasco is a prominent researcher working at the intersection of machine learning, computer vision, and humanoid robotics, with a particular focus on enabling autonomous robots to perceive, learn, and interact with the real world. His most influential contributions center on developing visual and tactile object recognition systems for humanoid platforms, most notably the iCub robot. Rosasco's work on deep convolutional neural networks for robotic object recognition has been foundational, demonstrating how robots can learn to identify objects from limited examples while achieving robust invariance to visual variability — research that has collectively garnered over 180 citations. His creation of the iCub World dataset introduced an innovative human-robot interaction paradigm for efficient data collection, addressing one of deep learning's most persistent challenges in robotics. Beyond vision, Rosasco has explored tactile exploration using Gaussian process classification for 3D shape reconstruction, and more recently expanded into whole-body motion planning and humanoid locomotion through learned trajectory generators. His 2018 survey questioning the maturity of object recognition from a robotics perspective reflects a thoughtful, critical voice in the field. Across his career, Rosasco has consistently bridged theoretical machine learning with practical robotic deployment, making substantial contributions to embodied artificial intelligence.

Research Focus

Key Achievements

Top Papers

1. 1
   Active perception: Building objects' models using tactile exploration

   58 citations · 2016
2. 2
   Object identification from few examples by improving the invariance of a Deep Convolutional Neural Network

   50 citations · 2016
3. 3
   Are we done with object recognition? The iCub robot’s perspective

   46 citations · 2018
4. 4
   Teaching iCub to recognize objects using deep convolutional neural networks

   35 citations · 2015
5. 5
   iCub World: Friendly Robots Help Building Good Vision Data-Sets

   30 citations · 2013
6. 6
   Interactive data collection for deep learning object detectors on humanoid robots

   28 citations · 2017
7. 7
   On-line object detection: a robotics challenge

   26 citations · 2019
8. 8
   Enabling Depth-Driven Visual Attention on the iCub Humanoid Robot: Instructions for Use and New Perspectives

   22 citations · 2016
9. 9
   Constrained DMPs for Feasible Skill Learning on Humanoid Robots

   21 citations · 2018
10. 10
    ADHERENT: Learning Human-like Trajectory Generators for Whole-body Control of Humanoid Robots

    17 citations · 2022

Key Collaborators

Contact & Links