Researchers in the U.S. and France have created robots that can adapt to damages in two minutes or less.
The results of the research leads the way for robots to be used in difficult settings like looking after elderly people, rescuing earthquake victims or doing household chores. Now, they can readily and easily cope up with damages that they could encounter in the real world.
Current robots are more strict in the way they work, as any part that doesn't work has to be replaced entirely so that the machine can accomplish its objectives.
While most self-learning systems take too long to find different options for continuing tasks since they look through billions of possibilities, the new system solves this problem with software that filters out options that don't work, BBC News reported.
Now, with what the authors call a "simulated childhood," the process is much quicker.
The six-legged robot uses a computer simulation of itself to create a map of every possible movement pattern it could do if faced with malfunction. Whichever new pattern that gives the fastest movement speed is what the robot then adapts to, making it possible for the machine to continue whatever task it is trying to accomplish.
The research team noted that this technique provides the robot with 10^47 behavior options, which is roughly equivalent to the number of molecules on Earth, Pacific Standard Magazine reported. The robot then predicts which choices would work best by testing each of them when the damage occurred. As a result, the robot would have fewer options when looking for a solution to a problem.
The researchers published their work in the journal Nature.
