Robotic singularity, here we come.
A team from the University of Southern Califonia created a robot that’s apparently so smart, it managed to teach itself how to walk without any previous programming. In fact, the robot isn’t even a complete robot – it’s a robotic limb that makes use of an algorithm named G2P, or ‘general to particular’.
Biomedical engineer Professor Francisco Valero-Cuevas says, “Nowadays, it takes the equivalent of months or years of training for a robot to be ready to interact with the world, but we want to achieve the quick learning and adaptations seen in nature.”
The team also claimed that the limb learned how to walk after just five minutes of letting it ‘play’ using random movements – something which they call ‘motor babbling’.
“These random movements of the leg allow the robot to build an internal map of its limb and its interactions with the environment,” says Ali Marjaninejad, one of the researchers. “If you let robots learn from relevant experience, then they will eventually find a solution that, once found, will be put to use and adapted as needed. The solution may not be perfect, but will be adopted if it is good enough for the situation.”
The robot even has personalized movements, thanks to the same ‘motor babbling’ process. Basically, the robots can use this to develop their own gait and stance, two things that are unique from one individual to another.
“You can recognize someone coming down the hall because they have a particular footfall,” Valero-Cuevas explains. “Our robot uses its limited experience to find a solution to a problem that then becomes its personalized habit – we get the dainty walker, the lazy walker, the champ… you name it.”
