'Psychic robot' predicts your intentions

Didn't mean to knock that cup of coffee over on your desk? That's OK. Your robot overlord will understand and maybe even keep you from doing it in the first place.

Michael Franco
Freelancer Michael Franco writes about the serious and silly sides of science and technology for CNET and other pixel and paper pubs. He's kept his fingers on the keyboard while owning a B&B in Amish country, managing an eco-resort in the Caribbean, sweating in Singapore, and rehydrating (with beer, of course) in Prague. E-mail Michael.
Michael Franco
2 min read

Enlarge Image
This device helped develop an algorithm that could predict our intentions. Justin Horowitz

For those worried about the singularity -- that time when computers or robots become smarter than us and start reproducing themselves -- new research out of the University of Illinois at Chicago might be a bit unsettling. Bioengineers there have developed a computer algorithm that can predict your intended actions, even if you accidentally change course.

"We call it a psychic robot," Justin Horowitz, a UIC graduate student research assistant and first author of a study appearing online in the journal PLOS One, said in a statement.

Machines that understand our intentions and override them when appropriate could help keep us safe. They could steer cars back on track in perilous conditions, for example. Or a smart prosthesis could counteract movements that might sabotage a stroke victim's intended movement.

The algorithm out of UIC could be a start. To develop it, Horowitz and his colleagues built a virtual desk. It consisted of a semi-transparent mirror that created the illusion that objects -- reflected from a display above the mirror -- were in the same space as the test subject's real hand, located below the mirror. While reaching for the objects, the test subject's hand was actually gripping the handle of a robot that could measure his movements and apply force, such as a slight nudge.

After the nudge, the robotic arm continued measuring the subject's hand movement but didn't otherwise interfere, Horowitz told CNET's Crave blog.

Using a comparison between nudged and undisturbed paths of movement, the researchers were able to arrive at their algorithm and design a computer model that could effectively predict the intent of the test subjects -- in this case, reaching for the virtual object. The UIC team found that it takes at least one-tenth of a second for a hand to adjust to a disturbance, such as a nudge or turbulence, but a computer is able to perform a calculation much faster than that.

"Say you're reaching for a piece of paper and your hand is bumped mid-reach, your eyes take time to adjust, your nerves take time to process what has happened, your brain takes time to process what has happened and even more time to get a new signal to your hand," said Horowitz, who worked with James Patton, UIC associate professor of bioengineering, on the study.

The notion of a helpful computer that knows us better than we know ourselves seems comforting. Then again, how much longer will it be till we ask our computer "helper" to do something for us, it overrides our intent and we hear, "I'm sorry Dave, I'm afraid I can't do that"?