Researchers in Europe have developed a new robot that uses algorithms to mimic human visual system for navigation.The machine is a wheeled platform controlled by a mechanic head, which has built in cameras to provide vision. Normally, the AI systems used for navigation build a map of its surroundings and update it as the robots move. The algorithm used on this project mainly consists of three components which are, human object recognition, motion estimation and decision making. The work brought together expertise from fields such as robotics, neuroscience and computer science and the project is named “decisions in motion”. The researchers note that the project goal is not necessarily surpassing the best autonomously navigating robot around, but rather to understand and animate real human motion decision making mechanism based on human visual processing. Also see the story on Technology Review.
The project consists of teams from many countires. The teams names and their tasks can be summarized as follows:
Karolinska Institutet, Stockholm, Sweden, which works on revealing human cortex motion decision making mechanism, Koninklijke Nederlandse Akademie van Wetenschappen, Amsterdam, Netherlands, works on visual cortex mechanisms that is used to distinguish an object from its background, Centre National de la Recherche Scientifique, Paris, France, works on visual computation strategies, University of Birmingham, Great Britain, works on neural systems that control visual search during motion, University of Regensburg, Germany, investigates the neural basis of self motion in humans, University of Ulm, Germany, models an artificial neural network process involved in motion segmentation, Fundacio Barcelona Media Universitat Pompeu Fabra, Barcelona, Spain, works on decision making, attention and learning mechanisms of the brain, Cambridge Research Systems Limited, Rochester, Great Britain, provides technology for investigations of the neural mechanisms, Scuola Superiore di Studi Universitare di Perfezionamento Sant’ Anna, Pisa, Italy, provides a robotic platform which is to perform like humans in similar settings and SpikeNet Technology, Labege, France provides algorithms based on human visual performance.
You can visit the project main page at http://www.decisionsinmotion.org/index.html