The robot subjects of the exhibit (via INRIA Flowers)
All the complexities of the human mind will take decades, if not centuries, to reproduce in robotics. (That is, barring some technological singularity.) Still, step by tiny step we are learning about what it would take to achieve artificial intelligence in robots. Pierre-Yves Oudeyer and his team from the INRIA ENSTA ParisTech are beginning to explore an artificial representation of one of the most intriguing and useful human traits: curiosity. The installation is called “Mathematics: A Beautiful Elsewhere”. Leading the project are Oudeyer, mathematician Mikahail Gromov and American director and visual artist, David Lynch. Oudeyer and his team designed the robots and their computational model for curiosity-driven learning. Gromov contributed his mathematical theory on ergo-systems and Lynch was responsible for designing the Ergo-robot habitat that looks like an egg and is mean to represent the “origin”.
To study artificial curiosity, they are equipping some intriguing looking robots with auditory and visual sensors to pique their interest and curiosity with visuals and sounds. The Ergo-robot’s software was inspired and designed to reproduce learning techniques used by human infants. Autonomously, the robot can detect sound, respond and hear feedback and engage in a loop interaction. Within these loops are data compressors to predict for the next input, along with reward value systems. The robots can learn to hear and emit new sounds from and to other robots and people. They even learn new behaviors. The team hopes the Ergo-robots will develop a form of communication amongst themselves and people as to eventually self-organize their own culture.
The display features 5 robots that are fastened to the ground at their base. Their bodies are comprised of a vertebrae-like mechanism that allows each robot to move like a snake sticking out of the ground. Their heads resemble prehistoric primate skulls and move around to look their environment, at each other and at people attending the display at the Cartier Pour l’Art Contermporain Foundation in Paris.
The installation will be open to the public until March 18, 2012. If these robots do start to learn and develop their artificial intelligence, it will be a good thing they are bolted to the ground.
To study artificial curiosity, they are equipping some intriguing looking robots with auditory and visual sensors to pique their interest and curiosity with visuals and sounds. The Ergo-robot’s software was inspired and designed to reproduce learning techniques used by human infants. Autonomously, the robot can detect sound, respond and hear feedback and engage in a loop interaction. Within these loops are data compressors to predict for the next input, along with reward value systems. The robots can learn to hear and emit new sounds from and to other robots and people. They even learn new behaviors. The team hopes the Ergo-robots will develop a form of communication amongst themselves and people as to eventually self-organize their own culture.
The display features 5 robots that are fastened to the ground at their base. Their bodies are comprised of a vertebrae-like mechanism that allows each robot to move like a snake sticking out of the ground. Their heads resemble prehistoric primate skulls and move around to look their environment, at each other and at people attending the display at the Cartier Pour l’Art Contermporain Foundation in Paris.
The installation will be open to the public until March 18, 2012. If these robots do start to learn and develop their artificial intelligence, it will be a good thing they are bolted to the ground.
Cabe
See the latest with our atificial friends in the Robotics group.
