Most of the news coming out of robotics research has us really worried about mankind’s future, but Cornell University finally brings us a glimmer of hope. Researchers there are working on developing an algorithm through physical feedback that will teach robots to be more careful with certain objects — like say when handling a sharp knife around highly stabbable humans.
The robot is free to first perform movements it assumes are ok, but a human operator can intervene at any point with the push of a button and then adjust the bot’s trajectory and movements to something a little safer. So, for example, if it grabs a knife and is unaware the long blade can cause significant damage, the human operator can move the robot’s arm to a safer position, and over time the algorithm will learn to be more careful when handling sharp objects.
The research isn’t just about protecting humans, though. A robot could be taught to handle fragile items like eggs or LCD screens with greater care, making them more effective in a factory setting. But on the flip side to all the good this research could do, is it really such a great idea to reinforce the idea for robots that knives and humans don’t mix? [Cornell University via TechCrunch]