Teaching Robots To Deceive

Researchers at the Georgia Institute of Technology, in Atlanta,Georgia have developed a few robots that deceive each other. 

How? By transferring the deceptive behavioral patterns of squirrels and birds into the small mobile robots. In other words, the research team developed algorithms that allow a robot to determine if it should deceive a human or other intelligent machine and techniques to help the robot select the best deceptive strategy and reduce its chance of being discovered.

Professor Ronald Arkin, a Regents Professor at the School of Interactive Computing at GeorgiaTech, and his team reviewed biological research results from squirrels showing how they gather acorns and store them in specific locations. Then, they patrol their hidden stashes, routinely going back and forth to check on them. When another squirrel shows up, the hoarding squirrel changes its behavior and instead of checking on the true location of its hidden stash, it visits empty stash sites, trying to deceive his competition.

The team implemented the same strategy into a robotic model and it took. The deceiving robot lured the competitor robot to the false locations, delaying the discovery of the protected resources. The hider robots were able to deceive the seeker robots in 75 percent of the trials.

Why you ask? According to Professor Arkin, this type of robot could be applied to the military or search and rescue operations in the future. Incidentally, the project is funded funded by the Office of Naval Research.

“This application could be used by robots guarding ammunition or supplies on the battlefield,” said Arkin. “If an enemy were present, the robot could change its patrolling strategies to deceive humans or another intelligent machine, buying time until reinforcements are able to arrive.”

Georgia Tech Regents professor Ronald Arkin (left) & research engineer Alan Wagner look on as the black robot deceives the red robot into thinking it's hiding down the left corridor. (Georgia Tech Photo: Gary Meek)

According to the research paper, a search and rescue robot may need to deceive in order to calm or receive cooperation from a panicking victim. Robots on the battlefield with the power of deception will be able to successfully hide and mislead the enemy to keep themselves and valuable information safe.

“Most social robots will probably rarely use deception, but it’s still an important tool in the robot’s interactive arsenal because robots that recognize the need for deception have advantages in terms of outcome compared to robots that do not recognize the need for deception,” said Alan Wagner, a research engineer at the Georgia Tech Research Institute and study’s co-author.

Looking to birds, the research team also created a simulation and demo based on birds that might bluff their way to safety. A bird known as an Arabian babblers that’s in danger of being attacked will join other birds and harass their predator. This mobbing process causes such a commotion that the predator will eventually give up the attack and leave.

Arkin’s team investigated whether a simulated babbler is more likely to survive if it fakes or feigns strength when it doesn’t exist. The team’s simulations, based on biological models of dishonesty and the handicap principle, show that deception is the best strategy when the addition of deceitful agents pushes the size of the group to the minimum level required to frustrate the predator enough for it to flee.

Arkin says the reward for deceit sometimes outweighs the risk of being caught.

“In military operations, a robot that’s threatened might feign the ability to combat adversaries without actually being able to effectively protect itself,” said Arkin. “Being honest about the robot’s abilities risks capture or destruction. Deception, if used at the right time in the right way, could possibly eliminate or minimize the threat.”

Via: "Forbes"