Isaac Asimov’s first law of robotics is well known amongst the Sci-Fi community. Something as powerful as, “A robot may not injure a human being” is pretty important, especially considering we’ve now got robots beating world-class human shogi players. But, apparently that first law doesn’t mean anything if robots don’t know the exact details as to how hard they can hit a human, before they injure a human. Thanks to a group of scientists, that law is currently being tested in the only way they see fit: by telling a robot to repeatedly punch humans.
Borut Povše, the lead scientist behind the research, has apparently obtained ethical approval from the University of Ljubljana, where the research was conducted. Povše has managed to convince six of his male colleagues to let the robot they created to repeatedly punch them in the arm, which is meant to determine human-robot pain thresholds. The idea behind the research is that, inevitably, a robot will collide with a human, in one way or another. The first stage of the research is to give robots a “definition” of human pain tolerance. The second, and just as important step, is for the robots to pull back on their movement, based on speed and acceleration in particular cases, so that if they do happen to come into proximity of a human, they can tone it down, and therefore cause minimal pain.
As Povše puts it, “Even robots designed to Asimov’s laws can collide with people. We are trying to make sure that when they do, the collision is not too powerful.” He goes on to add, “We are taking the first steps to defining the limits of speed and acceleration of robots, and the ideal size and shape of the tools they use, so they can safely interact with humans.” For the most part, it sounds like the research is designed to make sure that if a robot hits a human by accident, the result won’t be as catastrophic as it could be.
The robot, which was borrowed from Japanese technology firm Epson, was redesigned for the purpose of the research. The robot holds a tool in its grip, one that is blunt and round, and another that is “sharper,” and then repeatedly hits the human test subjects. The humans are then told to report their results. According to Povše, most of the colleagues judged the pain to be in the mild or moderate range. The tests will continue with a artificial human arm, to judge harder, and faster strikes against it.
The end result, hopefully, will be robots that are not capable of going past a certain set point in their design. Meaning, they will not be physically able to reach a speed, or an accelerated speed, which would be capable of injuring a human.
[via New Scientist]