In a laboratory at the University of Ljubljana in Slovenia, a robot is punching one of six men, repeatedly. This might even be happening right now, as you read these words. But don’t worry! When the pain is too much to bear, the victim will say so and the robot will stop. The experiment will be over.
Borut Povše, the (human) mind behind this project, isn’t flouting Asimov’s iconic laws, nor is he admininstering any kind of deliberate torture. As a matter of fact, his end goal is to keep people safe from robots, by precisely and thoroughly measuring the effects of human impacts and collisions with robots, alongside of which more and more factory workers will soon spend their days.
The human element of this experiment, in which six volunteers will report pain responses during sustained robot arm bombardment, is just the latest stage in an ongoing experiment–Povše has been investigating theoretical robot accidents for some time now. He kicked off his investigation due to what he sees as an inevitability of manufacturing. From his paper, “Cooperation of human operator and small industrial robot:”
Future development of industrial production performance and new technologies require coexistence of humans and robotic systems. Future robots will not work behind safety guards with locked doors or light barriers. Instead they will be working in close cooperation with humans which leads to fundamental concern of how to ensure safe physical human robot interaction
In earlier experiments, his methodology entailed striking a synthetic, sensor-packed arm with what he estimated to be realistic impacts–the kinds of accidents that might occur in a factory setting. The attacking arm was fitted with a pointy tip, with which it struck the rubber arm at various levels of force. The results weren’t terribly surprising: At a certain point, the robot arm exerted enough pressure on the arm that it would have caused tissue injury.
One goal of this early study, however, was to clear the way for similar experiments using actual humans, who can give subjective feedback about different levels of impact. In other words, they can provide an experiential “Yeah, that hurt!” to complement the “Impact energy density equals 2.52 J/cm^2″ calculations of the unfeeling arm apparatus. Industrial robots need to be able to work alongside humans without mangling us, and they also they need to be able to do so without causing routine discomfort.
Povše’s work will be among the first to address this somewhat cloudy area of robotics, and help establish a new, almost caring element in the relationship between robots and humans: A respect not just for bodily safety, but for pain. With his data, robot manufacturers may be able to tune smaller industrial robots to human pain tolerances, keeping the hybrid assembly lines of the near future not just safe, but comfortable.
