Team teaches robots to deceive

Posted by Emma Woollacott

Researchers have been playing hide-and-seek with robots to help them develop the ability to deceive. 

The Navy-funded research, from the Georgia Institute of Technology, is aimed at creating a robot that can deceive an enemy soldier by creating a false trail and hiding so that it won't be caught.

"We have developed algorithms that allow a robot to determine whether it should deceive a human or other intelligent machine, and we have designed techniques that help the robot select the best deceptive strategy to reduce its chance of being discovered," says professor Ronald Arkin.

The researchers' first step was to teach the robot how to recognize when deception was appropriate, using interdependence theory and game theory to develop algorithms that tested the value of deception in a specific situation.

A situation had to satisfy two key conditions to warrant deception - there must be conflict between the deceiving robot and the seeker, and the deceiver must benefit from the deception.

Then the robot was taught to create a false trail and hide. To test their algorithms, the researchers ran 20 hide-and-seek experiments with two autonomous robots. Colored markers were lined up along three potential pathways to places where the robot could hide. The hider robot randomly selected one of the three hiding locations and moved towards it, knocking down the markers along the way.

However, once it reached a point past the markers, the robot changed course and hid in one of the other two locations.

The hider robots were able to fool the seeker robots in 75 percent of trials, failing only when it didn't manage to knock over the right markers to create the false trail.

"The experimental results weren't perfect, but they demonstrated the learning and use of deception signals by real robots in a noisy environment," says co-author Alan Wagner.

"The results were also a preliminary indication that the techniques and algorithms described in the paper could be used to successfully produce deceptive behavior in a robot."

The researchers concede that there are ethical implications.

"We strongly encourage discussion about the appropriateness of deceptive robots to determine what, if any, regulations or guidelines should constrain the development of these systems," says Arkin.