Daniele Pucci

Researcher
Head of the Dynamic Interaction Control research Line

Contacts

Daniele.Pucci@iit.it
+39 010 71781 420

About

Education 

I received the bachelor and master degrees in Control Engineering with highest honors from "Sapienza", University of Rome, in 2007 and 2009, respectively. I also received the "Academic Excellence Award" from Sapienza in 2009. In 2013, I received the PhD title in Information and Communication Technologies from University of Nice Sophia Antipolis, with a thesis prepared at INRIA Sophia Antipolis, France, under the supervision of Tarek Hamel, and Claude Samson. The PhD program was jointly with Sapienza, university of Rome, so I also received a PhD in Control Engineering from “Sapienza".

 

Current position

Since August 2017, I am a research scientist leading the Dynamic Interaction Control lab at IIT. Our research can be decomposed into three research axes: bipedal-locomotion, physical-human-robot-interaction, and aerial-humanoid-robotics. The young, dynamic, and highly motivated team is composed of 25 persons.

 image1.jpeg

2015-2017, Senior postdoc, team leader of four PhD students  

From 2015 to August 2017 I have been a senior postdoc within the Dynamic Interaction Control Lab. I have beed coordinating four PhD students. 

Their domain of expertise is on planning, estimation, and control of floating base systems. Francesco Romano graduated in 2016, and is now a research engineer at Google DeepMind

          Francesco Romano             Yeshasvi Tirupachuri            Marie Charbonneau                  Gabriele Nava

         

 

 

2013-2015,  postdoc, in the coordination PI team of the CoDyCo project

From 2013 to 2015, I have been a  postdoc responsible for the experimental activities associated with the deliverables of CoDyCo, project number: 600716 FP7-ICT-2011.2.1 Cognitive Systems and Robotics. The project has been graded as excellent for four years. Given the success of the experimental activities obtained as an outocme of the project, I have been included in the coordination PI team of the project - see video below, minute 1:55 - coordinating a team of three PhD students.

 

 rovereto      

Interests

Control Robotics Mechanics

Projects

The research activities of the Dynamic Interaction Control lab can be decomposed into several research axes

Bipedal-locomotion

The goal of this research axis is to endow the humanoid robot iCub with a degree of bipedal locomotion. Prior to locomotion, the balancing capability is a fundamental property that must be studied and implemented on the robot. The video below show the balancing capacity of iCub. 

 

 While the following video shows improved balancing capibilities and preliminary bipedal locomotion of the humanoid robot iCub.

Physical-human-robot-interaction

Futuristic robots will be physically interacting with huaman beings. For this, it is fundamental to conceive control algorithms for humanoid robots interacting with a user. The video below shows some of the results in  physical-human-robot-interaction control. The user wears a sensorised suit retriving information used by the robot (human joint positions, velocities, and torques). The results were shown during the last year review meeting of the CoDyCo european project.


Aerial Humanoid Robotics

Some of the effort of my research group is on conceiving a new branch of Robotics implementing a degree of manipulation, and aerial and terrestrial locomotion. Among the infinity of platforms that may implement these three capacities, we have started to investigate the possibility of making a humanoid robot fly. The video below show preliminary results in aerial humanoid robotics. For further information, take a look at either the scientific paper or at the divulgative article that  IEEE Spectrum wrote about our work.

 In the following video, Richard Browning, test pilot of Gravity Industries, performs a demonstration at the IIT showing the engineering feasibility of the Aerial Humanoid Robotics

The PhD project 

During my PhD project, I mainly worked on three domains.

Aerial Robotics 

Feedback control of aerial vehicles in order to achieve some degree of autonomy remains an active research domain after decades of studies in the subject. The complexity of aerodynamic effects and the diversity of flying vehicles partly account for this continued
interest. The scientific community has dealt with the control of flying machines by mainly developing different strategies in relation to different classes of aerial vehicles – as exemplified by airliners and helicopters – and no unified approach has been developed so far. My research attempt contributes towards the development of such a unified approach by taking into account nonlinear aerodynamic effects in the control design.
 
The control framework assumes that the aerodynamic effects of rotational and unsteady motions are negligible, and that the means of actuation for an aerial vehicle consist of a body-fixed thrust force for translational motion and a control torque  for attitude monitoring. My research focuses on the guidance loop of the control problem. One of the main objectives has been to determine how to regulate the  thrust intensity and the vehicle orientation to compensate for the orientation-dependent external forces.
 
The following videos show the performances of the control approaches developed during my PhD. The objective was to stabilize the aircraft, which is subjected to aerodynamic forces that depend on the vehicle's orientation, about reference motions, specified by either a reference velocity or position. These simulations were performed with Matlab-Simulink. See below some of my results.
 

 

Wheeled Robotics 

Providing autonomous platforms that may improve the quality of our lives is an active research domain in robotics. Among these platforms, we studied the problem of a mobile robot that must follow a user. This robotic application can  be used for several tasks such as  shopping carts, butlers that may help for carrying heavy objects, or walking-aid systems. For such mobile platforms to work properly, two basic issues must be dealt with. The first one concerns the detection of the user and the estimation of its position with respect to the robot. The second issue is the design of feedback control laws that maintain a desired relative position between the robot and the user. I actively contributed to providing new solutions to these two issues. For instance, it is well known that position controllers typically used to stabilize the robot with respect to the user work well when the user is located in front of the robot and moves forward. In particular, the so-called jack-knife effect is problematic if the user starts moving backward. I proposed a solution that allows us to stabilize the robot w.r.t. the person with the latter located sideways – i.e. along the wheels’ axis – and also ensures jack-knife effect avoidance. It can be implemented with conventional sensor suite since the control law is position-based, i.e. the relative orientation of the person w.r.t. the robot is not needed to compute the control law. 

Nucleae Fusion

Accurate chamber pre-fills and plasma density control are primary concerns when dealing with tokamak machines, and I contributed to developing a control strategy for these two concerns. Plasma density evolutions are governed by a set of partial differential equations involving machine’s geometries, electromagnetic fields, atomic physics, and initial conditions of the entire system. This set of equations cannot in general be written in the state-space representation, and this renders almost inapplicable most of the classical  control methods. We proposed to take a different route by considering a linear, autonomous, model for the so-called averaged line plasma density. Then,  the control designed was performed on this model by applying a nonlinear PID with a classical anti-wind up. The chamber pre-fill control is achieved analogously. The whole system has been implemented using the real-rime framework MARTe, coded in C++.

 


Selected Publications

 

  1. L. Natale, C. Bartolozzi, D. Pucci, A. Wykowska, G. Metta "iCub: the not yet finished story of building a robot child", in press, Science Robotics
  2. D. Pucci, T. Hamel, P. Morin, and C. Samson. “Nonlinear Feedback Control of Axisymmetric Aerial Vehicles”. In: Automatica 53 (2015), pp. 72–78.
  3. D. Pucci, F. Romano, and F. Nori. “Collocated Adaptive Control of Underactuated Mechanical Systems”. In: IEEE Transactions on Robotics 31 (6 2015), pp. 1–10.
  4. D. Pucci, S. Traversaro, and F. Nori. “Momentum Control of an Underactuated Flying Humanoid Robot”. In: IEEE Robotics and Automation Letters 3.1 (2018), pp. 195–202. 
  5. F. Nori, S. Traversaro, J. Eljaik, F. Romano, A. Del Prete, and D. Pucci. “iCub Whole-body Control through Force Regulation on Rigid Noncoplanar Contacts”. In: Frontiers in Robotics and AI (2015).

Awards

 

2017 Finalist, Kuka innovation award, Co-aware team.

2015 Most interesting research topics, Premio Sapio per la ricerca e l’innovazine, Total submissions: 141, selected for publication on "DA, per la ricerca e l’innovazione". Title: "Tu, robot.". Online at http://www.daonline.info/pagine/45.pdf, p. 38.

2010 Vinci Program Grant, travel support grant, from “Université Franco Italienne”.

2009 Academic Excellence, from “Sapienza” University of Rome.

2004 Research project award, at the Triennial Electronics Exhibit, Technical Industrial State Institute “G. Vallauri”, Velletri (Italy).

1998 Tullio Fazi Prize, distinguished student award, from Rotary International 2080.