Plug And Play
So-called evolutionary roboticists raise the question: why go to the trouble of building a new and improved machine when current robots could do it for you?
The idea invokes a high-tech Darwinism, in which researchers’ ultimate goal is to design artificial intelligence and robots that can analyze their own source code and mate with others by combining bits and pieces of their code with that of other robots — to create offspring, much like organic life.
Just like biological life evolves to fill ecological niches, these robots’ offspring might be better adapted to their environments. And some research teams, Wired reports, are on their way to making this technology a reality.
Computer scientists at Vrije Universiteit Amsterdam built a simplified system that shows how future robots might swap and combine their “genetic” information.
Their recent research, published in the journal Nature Machine Intelligence and which involved programming two parent robots to code a new “offspring,” found that the resulting offspring contain a mixture of the parents’ code as well as some modules that seemed to have mutated or been blended on its own.
“It gives you a lot of diversity, and it gives you the power to explore areas of a design space that you wouldn’t normally go into,” David Howard, one of the scientists on the project, told Wired.
Twenty or so years down the road, Howard envisions that scientists could mass produce cheap robots that go out and attempt to perform a given task — and then “breed” those that are more successful into a new generation of robots that grow increasingly adept at that particular skill.
“One of the things that makes natural evolution powerful is the idea that it can really specialize a creature to an environment,” Howard told Wired.