Learning and Interaction Group at ADVR-IIT

Last update: 07/11/2011, Sylvain Calinon

The Learning and Interaction Group from the Department of Advanced Robotics at the Italian Institute of Technology (IIT) has been established by Sylvain Calinon in 2009.

Research statement

While accuracy and speed have for a long time been top of the agenda for robot design and control, the development of new actuators and control architectures is now bringing a new focus on passive and active compliance, energy optimization, human-robot collaboration, easy-to-use interfaces and safety.

The machine learning tools that have been developed for precise reproduction of reference trajectories need to be re-thought and adapted to these new challenges. For planning, storing, controlling, predicting or re-using motion data, the encoding of a robot skill goes beyond its representation as a single reference trajectory that needs to be tracked or set of points that needs to be reached. Instead, other sources of information need to be considered, such as the local variation and correlation in the movement. Also, most of the machine learning tools developed so far are decomposed into an offline model estimation phase and a retrieval/regression phase. Instead, learning in compliant robots should view demonstration and reproduction as an interlaced process that can combine both imitation and reinforcement learning strategies to incrementally refine the task.

The development of compliant robots brings new challenges in machine learning and physical human-robot interaction, by extending the skill transfer problem towards tasks involving force information, and towards systems capable of learning how to cope with various sources of perturbation introduced by the user and the task. We take the perspective that both the redundancy of the robot architecture AND the task can be exploited to adapt a learned movement to new situations, while at the same time improving safety and energy consumption. Through these new physical guidance capabilities, the robot becomes a tangible interface that can exploit the natural teaching tendency of the user (scaffolding, kinesthetic teaching, exaggeration of movements to highlight the relevant features, etc.).

Along with the learning aspects, such perspective also emphasizes the development of new interfaces capable of visualizing the learned skill in an interactive manner, in order for the user to assess the robot's progress, as well as estimating its current generalization capabilities and understanding of the task.

Toward these goals, together with my students and collaborators, I explore the following issues:

• Robust and compact representation of movements by the superposition of basis flow fields.
• Incremental learning of tasks by combining imitation and exploration strategies in a probabilistic framework.
• Safety embedded in the teaching mechanism by exploiting task redundancy and compliant control.
• Development of EM-based Reinforcement Learning strategies that can cope with real-world exploration trials, such as the use of multidimensional rewards and multi-resolution policies.
• User interfaces for the assessment of skills acquisition through active sensing and interactive data visualization.

The long-term view is to develop flexible learning tools that will anticipate the ongoing raise of compliant actuators technologies. In particular, the target is to ensure a smooth transition to passive compliant actuators and manipulators that can be safely used in the proximity of users, by considering physical contact and collaborative interaction as key elements in the transfer of skills.

Recent updates ADVR-IIT

25/10/2011	HSMM example added in the Sourcecodes section.
23/08/2011	Papers and videos for IROS'2011 added in the publications section and videos section.
09/11/2010	Advanced Robotics paper, Humanoids'2010 paper and Worskshop paper added in the publications section.
25/07/2010	It seems that people would not mind having a pancake flipping robot at home! At least for Saturday and Sunday mornings, as the video reached 100K view in one week-end and was featured on several blogs such as Wired, Robots.net, PopularScience, Boingboing, Gizmodo, Makezine, PlasticPals, RobotLiving, BotJunkie, TheDailyWh.at, TechnaBob, Neatorama, LaughingSquid, Geek.com, Ubergizmo, ZiggyTech, Photoxels, Eater.com, AllTop, MillionWordYear, or Cnet.com. Among the reactions (most of them related to robots taking over the world), we can read: I, for one, welcome our new pancake-flipping overlords. When robots take over the universe and turn humans into drooling minions, they will smell like pancakes. Why, you ask? Researchers (read: friends of evil robots) at the Italian Institute of Technology taught a giant robot arm to flip pancakes through reinforcement learning. Sure, it took the robot 50 tries just to get the first part down, but he's probably just playing dumb. Everyone knows you need an angular velocity of 36 degrees per second to flip a pancake!!! In the future, chefs and cooks will be replaced by robots. Who wouldn't want a MarioBataliBot9000 explaining how the sea and mountains influence Ligurian cuisine, all the while making dinner for four? It's hard not to laugh, watching this dumb bot flub flip after flip. But we won't be laughing when we're running for our lives, slowed down by a stomach full of fluffy pancakes. Robots cooking, feeding themselves, and feeding other robots. A disturbing trend?
22/07/2010	IROS'2010 paper (with pancake flipping robot), IROS'2010 paper (with ironing robot) and RAM article added in the publications section.
20/01/2010	ICRA'2010 paper added in the publications section.

Recent updates LASA-EPFL

14/10/2009	I am very pleased to announce that the book "Robot programming by demonstration: A probabilistic approach" has been published and is available for order.
14/10/2009	New video of iCub learning to dance.
04/08/2009	Chief Cook and Salvador DaBot have been presented at Campus-Party Valencia. Here are pictures and videos of the robot at CP Valencia.
16/07/2009	Chief Cook and Salvador DaBot have been presented at Campus-Party Bogota. Here are pictures and videos of the robot at CP Bogota.
15/07/2009	New paper for ICAR'09. New paper for Ro-Man'09. New video of HOAP-3 learning to handle a spoon to feed Robota by taking mashed potatoes from a plate. New video of WAM robotic arm learning to play ping-pong.
21/11/2008	There's a new video showing Salvador DaBot the portraitist robot here. You can also look at its first masterpieces here.
13/11/2008	Salvador DaBot (The Mustachioed Portraitist Robot v2.0) is being presented at the Google Zeitgeist Partner Forum.
12/11/2008	Three awards notifications arriving nearly the same week: Robotdalen Scientific Award (see the press release here and the interview here), ABB Award, and best paper award at Ro-Man'07. Cloud created through recurrence analysis of the words in my thesis (from Wordle.net)
24/05/2008	Our team won the First Prize in the Human-Robot Interaction Challenge at the International Conference on Robotics and Automation (ICRA'2008), where the Chief Cook Robot finished ex aequo with its friend Keepon!
23/05/2008	The Chief Cook Robot has been blogged on Otherfarm, Communist Robot, Engadget, New Scientist blog, Make blog, Hack a day, Dvice, BotJunkie, Machines like us, Hacked Gadgets, Technovelgy, and InternetActu (in French).
05/05/2008	Three new videos have been uploaded: The Chief Cook Robot (learning to prepare a cheese'n'ham omelet) Teaching two industrial robotic arms through a flexible scaffolding process Learning of a collaborative manipulation skill with the HRP2 robot
02/052008	The GMM-GMR sourcecode is now available in C/C++ (platform independent) on Sourceforge. The Matlab GMM-GMR sourcode is also available on Matlab Central File Exchange.
27/02/2008	New paper accepted for a workshop on Social Interaction with Intelligent Indoor Robots at ICRA'2008.
07/02/2008	The "humanoid robot learning like a child" appeared in a New Scientist video.
29/12/2007	The "humanoid robot drawing portrait" has been featured on several technology/robots blogs such as Gizmodo, Technabob, Botjunkie, Engadget, Neatorama or Crunchgear. The "humanoid robot learning like a child" also appeared on Botjunkie, Dvice and Hackedgadgets.
21/12/2007	A "humanoid robot finds learning childs play" article appeared in the Newscientist of this week.
12/12/2007	Research section updated.
23/09/2007	New sourcecode available in the sourcecodes section.
07/08/2007	The list of my publications is available here (Please wait until the whole list is loaded).
07/08/2007	The website is online!