Revolutionary ‘Psychic Robot’ Knows What You’re Doing Before You Do

Improving human lives has always been the motivation behind robotics. But a new mathematical algorithm developed in the United States may herald a new generation of robots that save human lives when brain injuries or other factors put those lives at risk.

Bioengineers at the University of Illinois at Chicago (UIC) say they’ve invented a way to read human intent and correct a wrong move in case of an interruption. A new paper published in the journal PLoS One suggests the new algorithm may be integrated into various robotic systems that are designed to assist humans in their mechanical movements. One key application would be new-generation smart prostheses.

“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,” Justin Horowitz, first author of the study, told UIC news service.

Horowitz and principal investigator James Patton, of the Rehabilitation Institute of Chicago, say their algorithm may be revolutionarily helpful in a wide range of conditions, from airplane pilots in conditions of heavy turbulence to stroke patients. For the latter, a smart prosthesis that incorporates the technology should be able to correct the intended trajectory of a move – like to bring a cup of tea to the mouth without spilling it – despite muscle spasms and tremors.

It works because the human nervous system is relatively slow on the uptake.

#AI with a predictive, trajectory inference engine yielding reaction times faster than humans. Yeah, clairvoyance. http://t.co/VvmjDX9hrI

— Mark Wyner (@markwyner) October 9, 2015

The research team’s work demonstrates that after a move has been disrupted, the intended trajectory remains in the brain for at least a tenth of a second.

“If intent can change following disturbance, it can only do so after some period of time due to processing and communication delays in the sensorimotor system that produces new motor commands,” the study explains.

That delay time may be utilized to develop and integrate a special filter into the most advanced robotic systems to make them capable of correcting unintended movement.

“This approach can take any ‘candidate model’ of the human arm and create a filter that estimates the intended trajectory,” the paper explains.

Imagine you’re driving down a straight road, and something makes you jerk the wheel to the left. The robot would recognize what you intended to do, and correct the swerve.

Unlike robots, humans are “slow, and because of that we have to have something that predicts the future,” Patton said.