Robots are very trustworthy. They don’t intentionally harm humans, unless they’re expressly told to do that as in the case of the recent sniper attack in Houston, Texas where the sniper was killed by a bomb-bearing robot. However, if a robot’s sensors can’t detect a human being coming to harm because of their actions, you end up with news articles labeled “Security Robot Violates First Law”. That’s what happened last Tuesday at Stanford Shopping Center in Palo Alto, California: a security robot encountered a 1-year-old child, couldn’t “see” it, and knocked the child down. Then, because it still couldn’t see it, it ran over the boy’s leg.
— Lilian Kim (@liliankim7) July 12, 2016
The 300-pound, 5-foot-tall “K5” mechanical guard was built by Knightscope, and functions mainly as a security camera on wheels. It alerts human security guards to disturbances and can detect known shoplifters with its camera. Unfortunately, the parents of the 16 month old child this one attacked say that their kid isn’t the first , pounced on an innocent 1-year-old at the Stanford Shopping Center, and the 16-month-old child’s parents said this isn’t the first time a child has found itself on the wrong side of the robot’s intentions.
“The robot hit my son’s head and then he fell down facing down,” Tiffany Teng, Harwin Cheng’s mother, told KGO. “The robot did not stop and it kept moving forward.” The robot machine rolled over the toddler’s right foot and scraped little Harwin’s left leg. While hardly life threatening, the little guy must have been terrified.
The K5 is very good at sensing adults a few feet away, but apparently isn’t so good at detecting much smaller humans that might be below its line of sight. This looks a lot less like a runaway robot with blood in its eyes and more like an engineering oversight.
Back to the drawing board, guys.