Scientists observed the process of evolution by natural selection in a new breed of "mother" robots that can design and test their own "babies." The results of these "motherly" tests are then used by the robot to create better-performing models in future generations, without the need for human intervention.
The incredible robot independently builds a number of "children," and tests which ones are most efficient, the University of Cambridge reported. These desirable traits are then passed down into future generations as the process continues. In the experiment, the mother robot created between one and five plastic cubes containing small motors. The robot then tested generations of 10 different offspring, and proved to have the ability to pass down the "fittest" traits through these generations.
"Natural selection is basically reproduction, assessment, reproduction, assessment and so on," said lead researcher Fumiya Iida of Cambridge's Department of Engineering, who worked in collaboration with researchers at ETH Zurich. "That's essentially what this robot is doing - we can actually watch the improvement and diversification of the species."
Each robot child contained a unique "genome" that was composed of a combination of between one and five different genes that commanded the robots' shape, construction, and motor commands. Evolution took place across the generations through "mutations" in which components of a gene were modified, genes were added or deleted, or a new gene was formed through the merging of genes between two individuals. Each child was tested on how far it could travel from one point to another in a given amount of time to determine which one was the fittest. The most successful individuals remained unchanged from generation to generation, while the least successful were subjected to mutations and gene crossovers. The researchers observed that robot performance improved as new generations were produced, and the average speed for the fastest individuals in the final generation was twice what it was in the first generation.
"One of the big questions in biology is how intelligence came about - we're using robotics to explore this mystery," Iida said. "We think of robots as performing repetitive tasks, and they're typically designed for mass production instead of mass [customization], but we want to see robots that are capable of innovation and creativity."
Most current work involving autonomous robots involves computer simulations that allow scientists to test up to millions of potential outcomes, but this new research saw the robot generate its own possibilities without the need for computer simulation. The disadvantage of this method is that each "baby" robot took about 10 minutes to create and test. In the future, scientists could use computer simulations to pre-select the fittest candidates, and use the real-world models for actual testing.
"It's still a long way to go before we'll have robots that look, act and think like us," Iida said. "But what we do have are a lot of enabling technologies that will help us import some aspects of biology to the engineering world."
The findings were reported in a recent edition of the journal PLOS One.