Who I am
I am a Postdoctoral Research in Robotics and AI. My research goal is to investigate and improve robots’ perception capabilities for autonomous manipulation.
6 facts about me
- Researcher in robotics and AI with four year experience.
- Proven C++ skills, with a good knowledge of YARP and iCub libraries.
- Tensorflow and Python knowledge for Deep Reinforcement Learning.
- Background: Robotics, Bayesian filtering, Optimization and Deep Reinforcement Learning.
- Daily research and work activities on Unix-based systems.
- Hardware experience with the humanoid robotic platform iCub.
- Main interests: Grasping, Localization, Perception, 3D object modeling and Deep Reinforcement Learning.
Current research activity
The goal of my research is to improve robots’ perception and manipulation skills by exploiting both visual and tactile information. The testing platforms for my research are both humanoid robots, such as the iCub and R1 robots, and industrial manipulators, like the Franka Emika Panda.
Recently, I got interested in Deep Reinforcement Learning and I’m currently leading a new team focused on Deep Reinforcement Learning for manipulation at the Humanoid Sensing and Perception team @ iCub Facility.
Skills
- Programming:
C++
(experienced),Python
(experienced),MATLAB
(prior experience) - Libraries,
YARP
(expert),IpOpt
(experienced),OpenCV
(basic knowledge) - OS:
Linux
(experienced),Windows
(prior experience) - Machine Learning Frameworks,
Tensorflow
(experienced) - Physics simulator,
Mujoco
(experienced). - Build and CI tools,
CMake
(basic knowledge) - Versioning Systems,
Git
(experienced). - Robots & Platforms:
iCub
- Soft skills:
Time management
,Problem solving
,Critical thinking
,Teamwork
,Decision Making
,Motivation
Topics of Interest
Grasping
Manipulation
Object Localization
Perception
Deep Reinforcement Learning
3D object modeling
PhD in Advanced and Humanoid Robotics @ Istituto Italiano di Tecnologia
I got my PhD in Advanced and Humanoid Robotics on April 2019 with the dissertation of the Thesis entitled Sense, Think, Grasp: A study on visual and tactile information processing for autonomous manopulation, where I proposed possible solutions to some key components of autonomous manipulation, focusing in particular on the perception problem and testing the developed approaches on the humanoid robotic platform iCub. The main contributions of my PhD thesis consists of:
- an algorithm for modeling and grasping unknown objects with superquadrics function;
- a pipeline for executing the handover task with the iCub humanoid robot;
- a tactile localization and recognition algorithm, based on the Unscented Particle Filter;
- a novel exploration strategy for Deep Reinforcement Learning algorithms.
Research Experience
Here is a summary of my experiences in research field (click to expand
):
Visiting Scholar @ BAIR, UC Berkeley (January - July 2018)
My research activity at Bair focuses on the design of new deep reinforcement learning techniques aimed at improving robot manipulation and grasping capabilities.
Easy Peasy Robotics Mentor @ Campus Party Italia (2017)
I was one of the mentors and organizers of Easy Peasy Robotics, a 2-days crash course whose aim was to provide participants with a brief overview of the research problems and applications related to humanoid robot programming, from perception to control. An interview (in Italian) about Campus Party experience is available here.
Europen Project TacMan: Tactile Manipulation (2017)
TacMan is a project founded by the European Union, FP7 ICT Cognitive System and Robotics, no. 610967. My work for the TacMan project contributed to improving recognition and manipulation skills for the humanoid robot iCub. I developed a model-based tactile object localization and recognition algorithm and a novel pipeline in order to make the iCub robot perform the handover task, i.e. transfer an object from one hand to the other. A video of successful handovers is available here.
Brains, Minds and Machines Summer School (BMM 2016)
BMM summer school is organized by Harvard Medical School September, and Massachusetts Institute of Technology, Woods Hole, Massachusetts, US. An intensive three-week course gives advanced students a “deep end” introduction to the problem of intelligence – how the brain produces intelligent behavior and how we may be able to replicate intelligence in machines. The summer school selection process is very competitive due to the increasing number of applications and the small number of available positions. In 2016, 30 students have been selected among 300 nearly. The school requires the accomplishment of a 3 week project, for which I implemented an algorithm, capable of detecting and recognize activities in real videos. I achieved my goal by modeling the problem through Hidden Markov Models and by using Bayesian Regression as main approach.
International Computer Vision Summer School (ICVSS 2016)
The International Computer Vision Summer School is organized by University of Cambridge and University of Catania, Italy. The tenth edition of ICVSS provided both an objective and clear overview and an in-depth analysis of the state-of-the-art research in Computer Vision. The courses were delivered by world renowned experts in the field, from both academia and industry, and covered both theoretical and practical aspects of real Computer Vision problems as well as examples of their successful commercialisation. (Selected students: 150/396.)
Research Fellow @ iCub Facility (2015)
I have partnered with the Italian Institute of Technology during my M.Sc. thesis, about 6D object tactile localization, i.e. the problem to estimate the 6-DOF pose of a tridimensional object, whose model is known, by using the tactile measurements collected with the robot iCub.
The iCub Summer School - Veni Vidi Vici (VVV 2015)
The school focused on humanoid robotics, with the goal to foster collaboration on robot software across the boundaries and lifetimes of specific platforms and projects.
52th IEEE Conference on Decision and Control (CDC 2013)
During my M.Sc course I joined the IEEE Conference on Decision and Control as a Crew Member. The CDC is recognized as the premier scientific and engineering conference dedicated to the advancement of the theory and practice of systems and control.
Education
- M. Sc. with honours in Electrical and Automation Engineering (GPA 4.0/4.0), University of Florence, Italy, 2013 - 2015.
- B. Sc. with hounours in Electronic and Telecommunication Engineering (GPA 3.96/4.0), University of Florence, Italy, 2010 - 2013.
Awards
- RAS Travel Grant, at IEEE International Conference on Robotics and Automation (ICRA), Singapore, June 2017.
- Dr. Kanako Miura Travel Support Award, at IEEE International Conference on Humanoids Robotics, Cancun, Mexico, Novmber 2016.
- AEIT Renato Marian Award, as best student graduated in Information Engineering, Florence, Italy.