CARME TORRAS GENÍS

CARME TORRAS GENÍS

ERC Project: CLOTHILDE: Cloth manipulation learning from demonstration

Prof. Carme Torras aims to develop a theory of cloth manipulation, implementing a prototype in the laboratory for three applications: recognising and folding clothing, putting an elasticated cover over a mattress or a car seat and helping old and/or disabled people to get dressed.
The project aims to set the basis for versatile manipulation of cloth by robots. The theoretical basis will be to link automatic learning and computational topology methods to produce a theory of manipulated cloth deformation that leads to a general framework so that robots can learn to manipulate clothes from human demonstrations. This framework will cover non-expert teaching of a task, perception of robots and learning skills, task-based representation of cloth, probabilistic planning, performing robot tasks from different initial conditions, fault diagnosis and informed requests for human help.

Bio

My first dilemma as a student was having to choose between sciences and arts for Baccalaureate. How could I cheerfully walk away from such a large chunk of knowledge? I still think that this quandary, maybe even more acute these days, is dragging down the educational system in this country. I got over it at university by enrolling in both Mathematics at the UB and Philosophy and the Arts at the UNED (Open University).

Above all, I owe my determination at this point in my life to three people: my father, who was ahead of his time and really believed in what women could achieve; my mother’s brother who spent Saturday afternoons with me doing fun mathematical puzzles, and a broad-minded Baccalaureate teacher. They all encouraged me to never give up on anything.

I was intrigued by the processes for shaping thought, intelligence, emotions… in short, the brain. After finishing my Maths degree, for the year when I worked for an IT firm, I read a few books on Artificial Intelligence: (AI) – an upcoming discipline at the time – written by well-known researchers such as Rosenblatt, Minsky, Feigenbaum, Newell and Simon, that really made an impression on me. One was decisive in my scientific career: “Brains, machines and mathematics” by Michael A. Arbib. I wrote to him to say that I would love to study what he explained in his book on how to model the brain with computers and to my great surprise, he wrote back and invited me to study the Computational Sciences Master’s degree, specialising in Brain Theory, at the University of Massachusetts. I applied for a Fulbright scholarship and had to quite literally dash over there at the last minute. I was spending the summer in Vic at the time and we didn’t have a phone; I received a telegram telling me that I had been given the scholarship and that they were expecting me in Massachusetts in three days.

When I finished my master’s degree, I faced my second dilemma: return to Barcelona or stay in the USA. I prioritised my personal life over academia and, as I could not find anywhere in Catalonia to continue my research on computational models in neurosciences, I moved into robotics, where I was able to apply biologically-inspired models to perception, planning, reasoning, movement control and above all, to automatic learning implemented by robots.

At the Institute of Robotics and Industrial IT, a mixed CSIC and UPC centre, I have the privilege of leading a magnificent research group on Robotised Perception and Manipulation. I have supervised eighteen doctoral dissertations and I am proud of my PhD students who are now outstanding researchers in prestigious centres such as the EPFL, the University of Canberra, UFPR and of course the CSIC; some also hold important positions in leading technology companies.

My website (http://www.iri.upc.edu/people/torras) compiles my scientific production and the recognition I have received. I would also like to highlight my interest in the social implications of robotics and AI. Convinced that science fiction can stimulate the ethical debate on the applications of these new technologies, I have written novels entitled “La mutación sentimental” (Editorial Milenio, 2012) —which won the Pedrolo and Ictineu awards — and “Enxarxats” (Males Herbes, 2017). The former was translated into English as “The Vestigial Heart” (MIT Press, 2018) and published alongside online materials as part of a course on Ethics in Social Robotics and Artificial Intelligence (http://mitpress.mit.edu/books/vestigial-heart).

More about Carme Torras Genís

Interview for Magazine Digital
Sense Ficció TV3

Bio

My first dilemma as a student was having to choose between sciences and arts for Baccalaureate. How could I cheerfully walk away from such a large chunk of knowledge? I still think that this quandary, maybe even more acute these days, is dragging down the educational system in this country. I got over it at university by enrolling in both Mathematics at the UB and Philosophy and the Arts at the UNED (Open University).

Above all, I owe my determination at this point in my life to three people: my father, who was ahead of his time and really believed in what women could achieve; my mother’s brother who spent Saturday afternoons with me doing fun mathematical puzzles, and a broad-minded Baccalaureate teacher. They all encouraged me to never give up on anything.

I was intrigued by the processes for shaping thought, intelligence, emotions… in short, the brain. After finishing my Maths degree, for the year when I worked for an IT firm, I read a few books on Artificial Intelligence: (AI) – an upcoming discipline at the time – written by well-known researchers such as Rosenblatt, Minsky, Feigenbaum, Newell and Simon, that really made an impression on me. One was decisive in my scientific career: “Brains, machines and mathematics” by Michael A. Arbib. I wrote to him to say that I would love to study what he explained in his book on how to model the brain with computers and to my great surprise, he wrote back and invited me to study the Computational Sciences Master’s degree, specialising in Brain Theory, at the University of Massachusetts. I applied for a Fulbright scholarship and had to quite literally dash over there at the last minute. I was spending the summer in Vic at the time and we didn’t have a phone; I received a telegram telling me that I had been given the scholarship and that they were expecting me in Massachusetts in three days.

When I finished my master’s degree, I faced my second dilemma: return to Barcelona or stay in the USA. I prioritised my personal life over academia and, as I could not find anywhere in Catalonia to continue my research on computational models in neurosciences, I moved into robotics, where I was able to apply biologically-inspired models to perception, planning, reasoning, movement control and above all, to automatic learning implemented by robots.

At the Institute of Robotics and Industrial IT, a mixed CSIC and UPC centre, I have the privilege of leading a magnificent research group on Robotised Perception and Manipulation. I have supervised eighteen doctoral dissertations and I am proud of my PhD students who are now outstanding researchers in prestigious centres such as the EPFL, the University of Canberra, UFPR and of course the CSIC; some also hold important positions in leading technology companies.

My website (http://www.iri.upc.edu/people/torras) compiles my scientific production and the recognition I have received. I would also like to highlight my interest in the social implications of robotics and AI. Convinced that science fiction can stimulate the ethical debate on the applications of these new technologies, I have written novels entitled “La mutación sentimental” (Editorial Milenio, 2012) —which won the Pedrolo and Ictineu awards — and “Enxarxats” (Males Herbes, 2017). The former was translated into English as “The Vestigial Heart” (MIT Press, 2018) and published alongside online materials as part of a course on Ethics in Social Robotics and Artificial Intelligence (http://mitpress.mit.edu/books/vestigial-heart).

ERC Project: CLOTHILDE: Cloth manipulation learning from demonstration

Prof. Carme Torras aims to develop a theory of cloth manipulation, implementing a prototype in the laboratory for three applications: recognising and folding clothing, putting an elasticated cover over a mattress or a car seat and helping old and/or disabled people to get dressed.
The project aims to set the basis for versatile manipulation of cloth by robots. The theoretical basis will be to link automatic learning and computational topology methods to produce a theory of manipulated cloth deformation that leads to a general framework so that robots can learn to manipulate clothes from human demonstrations. This framework will cover non-expert teaching of a task, perception of robots and learning skills, task-based representation of cloth, probabilistic planning, performing robot tasks from different initial conditions, fault diagnosis and informed requests for human help.

More about Carme Torras Genís

Interview for Magazine Digital
Sense Ficció TV3