From the moment Boston Dynamics cemented the ‘robot dog’ in our collective consciousness, we’ve thoroughly enjoyed watching them get pushed and kicked around. There is an undeniable sense of relief in acknowledging robotic clumsiness.
But things are changing as these robots get more and more adept at picking themselves back up. A team of researchers from China’s Zhejiang University and the University of Edinburgh have designed a software that teaches a test subject robot dog how to respond to complex and unanticipated external stimuli.