This release is available in Spanish.
Members of the Biomedical Engineering research team of the UPNA-Public University of Navarre are working on the project "Universalisation of computer interfaces based on eye tracking," which aims to develop a device that costs less, has good resolution and greater precision than those achieved to date with systems of this type. The project, funded by the Spanish Ministry for the Economy and Competiveness, is being run in collaboration with the IT University of Copenhagen. Its three-year execution period terminates at the end of 2012, although a new line of research has been opened up to extend the eye tracking system outdoors.
Basically, eye tracking is a process that enables the point on which the gaze is fixed (where we are looking) to be gauged. As Arantxa Villanueva-Larre, the head researcher on the project, points out, until now the applications of eye tracking systems have been rather limited and have mainly concentrated on helping people with disabilities: eye tracking enables the computer keyboard and mouse to be substituted by a system that recognises which part of the monitor we are looking at and, either by means of a wink or staring for a time, to interact with the computer cursor. "It is used for people with ALS (Amyotrophic lateral sclerosis), people mentally capable but who cannot move, since it would appear that movement controlled by the eyes is still maintained in the advanced phases of the disease," she explains.
Eye tracking is also used in market research (to see what products the person looks at when they go round a supermarket, to know how a web page is read and which points attract people's attention, etc.). "In recent years there has been more widespread development of this technology and the range is opening out to see other possibilities: from applications for videogames to outdoor systems, like assistance in driving vehicles, for example."
Lower cost, more precision
There are eye tracking systems already being marketed by various companies with a very specific hardware and which achieve results of great precision, but their prices are so high and their operating requirements so strict that applying them in a domestic environment is difficult. "The idea behind our project," as Prof. Villanueva points out, "is to come up with eye tracking at a low cost, and in such a way that the user can install the software in his or her device and use a webcam for the purpose, without needing additional illuminators or very expensive optical components."
The UPNA research team working on the project is made up of Arantxa Villanueva-Larre, Rafael Cabeza-Laguna, Sonia Porta-Cuellar, Juan José Cerrolaza-Martínez, Laura Sesma-Sánchez, Mikel Ariz-Galilea, Leonardo De Maeztu-Reinares and Victoria Ponz-Sarvise. The problems they have come up against are in fact related to the optical components and the image capture devices. "When you work with an ordinary webcam, the quality of the image is not that good and detecting the pupil is trickier. For example, we have practically calculated the centre of the iris already, but a series of operations and estimates have to be redone to be able to specify which point the user is looking at on the screen."
Within the framework of the project they have produced a database of over one hundred users, who have been recorded while looking at different points; the data will be used as a test bench and reference. "We also have to develop various algorithms and detect what influence the position of the head and angles of vision have, because these results will be crucial when achieving a precise tracking of the gaze. For example, in algorithms detecting the corners of the eyes we have achieved good results for the outer corner of the eye, although the results are not as good when we try to detect the inner corner."
What is more, in the course of the research the project team has set itself a new goal: to extend the eye tracking system outdoors and study its potential application in different environments, like inside a vehicle. "We have done tests on videos recorded outside and the results obtained are encouraging us to continue along this line. In the case of vehicles, one of the biggest difficulties is the continual change in outdoor illumination, a key element when it comes to analysing images. Nevertheless, we have managed to fix the centre of the iris, but we still have to find out what point the eye is looking at."
The researchers believe that if this line were to yield satisfactory results, new possibilities would open up: "Imagine you are driving and during an unspecified number of images the system does not detect your eyes. We could then get an alarm to activate because we would understand that you are not looking where you are going or something has happened to you. We are considering applications of this kind in systems that will enable us to accurately detect what part of the vehicle or exterior spot you are looking at and then to use that information."