Saturday, February 24, 2007

Robots almost humans.

http://www.youtube.com/watch?v=caty27Fe1_o


The Dra. Lola Cañamero of the University of Hertfordshire, leader of Feelix Growing Project has before her a great project : to build robots that learning from humans will be capable of being related appropriately social and emotionally with them. By suddenly it is studied to condition in robots in development, the imprinting (baby robots that follow and imitate the first object that they see like human mothers). Feelix Growing, is a project headed by british that involves 6 countries, 25 robotists, development psychologists and neuroscientifics. According to Cañamero, the learning was given by means of emotional and tactile feedbacks received from humans, with positive reinforcements (kind words, correct conduct, aid to robots, etc), processed by videocameras, audios, sensors that narrow the distance among robots and humans. Initially facial bosses and human movements will be prioritized, adapted to neural artifical networks gotten into focus in baby robots that grow and help humans in their daily works. It is expected that artificial neural networks be capable of being adapted to constant changes, detecting human inputs of conduct, voice, movement, in which the observational learning adopted initially stunning emotions in robots (ire, happiness, solitude). Other participants in this project are: Center National of the Recherche Scientifique, Universite of Cergy Pontoise, Ecole Polytechnique Federate of Lausanne, University of Portsmouth, Institute of Communication and Computer Systems, Greece, Entertainment Robotics, Denmark and BE Aldebaran Robotics, France.

We believe that the aproximation is wrong and alone will serve to do but visible the conceptual error, because splits of an imitative external base and not from internal, own robot development. For us the interaction with humans and the environment should be given from a long-term plan that consider as a great goal the recognition on the part of the robots its own internal ego, that permit to differentiate the good thing of the evil in a general sense by their own, effect to be reached when the robots will possess powerful multimedia mechanisms in their CPUs. Then the emotions will be alone good or bad things, collateral effect.

Robots casi humanos.

La Dra. Lola Cañamero de la Universidad de Hertfordshire, lider de Feelix Growing tiene ante si un proyecto de polendas : construir robots que aprendiendo de humanos sean capaces de relacionarse apropiadamente social y emocionalmente con ellos. Por de pronto se estudia condicionar en robots en desarrollo, el imprinting (robots bebes que siguen y imitan primer objeto que ven (madres humanas). Feelix Growing, un projecto liderado por britanicos involucra a 6 paises, 25 robotistas, psicólogos desarrollistas y neurocientificos. Segun Cañamero, el aprendizaje se dara mediante feedbacks emocionales y tactiles recibidos de humanos, con refuerzos positivos (palabras amables, conducta correcta, ayuda al robot,etc), viabilizadas por videocamaras, audio, sensores acortadores de la distancia entre la maquina y los humanos. Inicialmente se priorizaran patrones faciales y de movimiento, adaptados a redes neuronales artifíciales focalizadas en robots bebes que crecen y ayudan a humanos en sus labores diarias. Se espera que las redes neurales artificiales sean capaces de adaptarse a cambios constantes, detectando patrones ingresantes humanos de conducta, voz, movimiento, en los que el aprendizaje observacional adoptará inicialmente las emociones más impactantes (cólera, felicidad, soledad), en la conducta de los robots .Otros participantes en este proyecto son: Centre National de la Recherche Scientifique, Universite de Cergy Pontoise, Ecole Polytechnique Federale de Lausanne, University of Portsmouth, Institute of Communication and Computer Systems, Greece, Entertainment Robotics, Denmark and SAS Aldebaran Robotics, France. Creemos que la aproximacion es equivocada y solo servira para hacer mas visible el error conceptual, porque parte de una base externa imitativa y no interna de desarrollo propio.
Para nosotros la interacción con humanos y el medio ambiente debe darse a partir de un plan a largo plazo que tenga por objetivo el reconocimiento por parte de los robots de su propio yo interno, que le permita diferenciar lo bueno de lo malo en un sentido general, efecto a alcanzar cuando los robots posean poderosos mecanismos de efectos multimedia en sus CPUs. Entonces las emociones serán solo un efecto colateral de lo bueno y lo malo.

No comments: