L. Natale(1), F.Nori(2), U. Pattacini(1), V. Tikhanoff(1), M. Randazzo(1), G. Metta(1)
(1) iCub Facility
(2) Robotics, Brain and Cognitive Sciences
Istituto Italiano di Tecnologia (IIT)
Via Morego, 30 - 16163 Genoa, Italy
The iCub is a humanoid robot shaped as a four years old child. It is available as an open system platform following the GPL license. The iCub was originally designed by a consortium of 11 partners, with background ranging from engineering to neurophysiology and developmental psychology, within the RobotCub Integrated Project funded by European Commission through its Cognitive Systems and Robotics Unit. The iCub can crawl on all fours, sit up and walk. Its hands allow dexterous manipulation and its head and eyes are fully articulated. It has visual, vestibular, auditory, and tactile sensory capabilities. The low-level software allows reading information from the sensory system and controlling the position or the force at the joints.
In the past years large efforts have been devoted to the development of the software infrastructure (the YARP middleware), tools, libraries and simulators that facilitate and support development and integration of complex behaviors. Particular attention has been devoted to provide a simple but powerful infrastructure with a gentle learning curve that accommodates the requirements of different users (experienced roboticists, or researchers from diverse fields like machine learning, computer vision or cognitive science). The iCub platform has recently evolved to incorporate whole-body distributed tactile and force/torque sensing, series elastic actuators for compliant walking experiments. This has boosted the development of new libraries and simulators to precise model and control interaction forces on the whole-body and study dynamic tasks.
This tutorial session will feature a series of talks aimed to illustrate with practical examples the software tools for connecting to the robot, reading the sensors and controlling the actuators in different modalities (joint, task space, position or torque). We will also describe the latest development that includes software simulators, libraries for machine learning and whole-body control.
The tutorial session has the following goals. The first is to illustrate the software architecture of a full humanoid robot, including the interface to all the available sensors and the actuators. The second goal is to disseminate various software tools that have been developed in the past few years, some of which have been described only partially in scientific publications. All the software is open-source and it can be freely used and adapted to other research projects and robots. Finally, the tutorial session will provide an overview of the latest software development on the iCub platform. An iCub will be available at the tutorial during a hands-on session to give participants the possibility to write and test code on the real robot.
This tutorial session goes well beyond the users of the iCub and covers a broader audience of robotic researchers. The session will provide the participants with the unique opportunity to use a complex robotic system and get an overview of its software infrastructure. The software libraries and tools that will be illustrated in the tutorial are in fact open-source and generic and can be used by students and researchers in their own projects. The simulators can be used by researchers but also for training activities in classes.
|9:00||An introduction to the iCub robot||Giorgio Metta - Istituto Italiano di Tecnologia|
|9:10||Communication and coordination using the YARP middleware.||Lorenzo Natale, Ali Paikan - Istituto Italiano di Tecnologia, Italy|
|9:33||A software library for whole-body control.||Francesco Nori and Silvio Traversaro - Istituto Italiano di Tecnologia, Italy|
|9:56||GURLS: A Least Squares Library for Supervised Learning.||Alessandro Rudi and Lorenzo Rosasco - University of Genoa, Italy|
|10:21||YARP Plugins for Gazebo Simulator: development and application on the iCub and COMAN robots||Alessio Rocchi, Enrico Mingo, Silvio Traversaro - Istituto Italiano di Tecnologia, Italy|
|11:05||Robotran: A Fast Symbolic, Dynamic Simulator interfaced with Yarp||Timothee Habra - Université Catholique de Louvain and Houman Dallali - Istituto Italiano di Tecnologia, Italy|
|11:38||iCub interacting with humans: software tools and best practices||Serena Ivaldi - INRIA, France|
|12:01||The Modular Behavioral Environment (MoBeE): Reactive Collision Avoidance and Offline Motion Planning Under a Single Software Framework||Mikhail Alexander Frank - IDSIA, Switzerland|
|12:24||Modelling Software Systems in Experimental Robotics for Improved Reproducibility - A Case Study with the iCub Humanoid Robot||Florian Lier, Sven Wachsmuth, Sebastian Wrede - Research Institute for Cognition and Robotics (CoR-Lab), Bielefeld University, Germany|
|14:30||During the afternoon session participants will have the possibility to view live demos and experiment with the iCub or the simulator with hands-on exercises||Ugo Pattacini, Alessandro Roncone, Vadim Tikhanoff, Marco Randazzo, Alessio Rocchi, Enrico Mingo - Istituto Italiano di Tecnologia, Italy|
|17:30||End of day|
We acknowledge financial support by the European Projects Xperience (FP7-ICT-270273), Walkman (FP7-ICT-611832), CodyCo (FP7-ICT-600716), KoroiBot (FP7-ICT-611909), Tacman (FP7-ICT-610967) and WYSIWYD (FP7-ICT-612139).