People who have lost the use of their body because of multiple sclerosis or spinal cord injury, for example, can usually move their eyes, because eyes are directly connected to the brain. Several technologies allow people to stare at arrows on a computer and direct the movement of a wheelchair, but there is a considerable delay between the movement of the eyes and the chair, and the person can't look around while manipulating the chair.
To overcome this problem, Aldo Faisal at Imperial College London and his colleagues have developed software that uses subtle eye movements to distinguish when a person is looking around and when they want to move. The team was due to showcase the technology at the Imperial Festival in London at the weekend.
"Current eye tracking software doesn't allow you to look around while moving," says Faisal. "And technologies that use brainwaves to control wheelchairs aren't common because it takes many months to train a person to use them, and then you need to really concentrate to move – it's not natural."
His team has observed how people move their eyes when walking around and used the data to build software that decodes a person's intentions. The finished system involves two cameras – one trained on each eye – that observe eye movements and pass that information onto a laptop computer, which works out which direction and how far into the distance a person is looking.
But then you've got the King Midas problem, says Faisal. "Everything he touched turned into gold, and we don't want to move everywhere we look."
Exactly how they solved this problem is still under wraps, he says, but it involved analysing subtle eye movement patterns to distinguish those relevant to locomotion from those we use when merely looking around. "Our software can tell the difference between looking at someone using a coffee machine, and wanting to walk over to that coffee machine," says Faisal.
The system responds within 10 milliseconds to a person's intention to move. Typically anything under about 15 or 20 milliseconds feels instantaneous, says Faisal.
The team has tested the system on people without physical disabilities and found that they were able to steer through a crowded building faster and with fewer mistakes than with current technologies that track eye movements.
Faisal says the team hopes to have the system ready for sale within three years. If successful, it could be adapted for other uses, like piloting a drone or a plane. It could also work in advertising to get people's attention as they are window shopping, say. If the storefront could watch your eyes and tell from the pattern of your eye movement that you are about to walk away, it might change and give you another advertisement to keep you interested.
The whole system is likely to cost about £50. "Our technology can be crap and cheap because all the smartness is in the software," says Faisal.
Syndicated content: Helen Thomson, New Scientist
Image: Rowan Hooper