A new robotic system that can evolve to optimise its performance could provide a blueprint for creating robots able to adapt to their environment.
Researchers from the University of Cambridge and ETH Zurich created a ‘mother’ robot arm that is able to design, build and test 'baby' robots before using the results to improve the performance of the next generation by retaining useful traits and discarding disadvantageous ones.
The entire system operates without human intervention or computer simulation beyond an initial command to build a robot capable of movement at the fastest speed possible from between one and five plastic cubes with a small motor inside.
The team conducted five experiments, in which ten generations of ten children were created by the mother robot, and the 'fittest' individual in the final generation moved twice the distance of the first before its power ran out.
“Natural selection is basically reproduction, assessment, reproduction, assessment and so on,” said lead researcher Dr Fumiya Iida, from Cambridge. “That’s essentially what this robot is doing – we can actually watch the improvement and diversification of the species.”
In a paper published in the open access journal PLOS One today, the researchers describe how each baby robot has a unique ‘genome’ made up of a combination of between one and five different genes, that contains all of the information about the child’s shape, construction and motor commands.
The genomes of the fastest individuals in each generation remained unchanged in the next generation, while those of the less successful children were subjected to 'mutation’ - where components of one gene are modified or single genes are added or deleted - and ‘crossover’ - where genes from two individuals are merged.
“One of the big questions in biology is how intelligence came about – we’re using robotics to explore this mystery,” said Iida. “We think of robots as performing repetitive tasks, and they’re typically designed for mass production instead of mass customisation, but we want to see robots that are capable of innovation and creativity.”
Most of the work in the growing field of evolutionary robotics is done using computer simulation as they can create thousands of possibilities in a short amount of time, but the Cambridge team says there is often a mismatch between simulated and real-world behaviour.
With each baby robot taken roughly 10 minutes to design, build and test, however, Iida says in future the team may use a computer simulation to pre-select the most promising candidates before using real-world models for actual testing.