London - A robot has taught itself the principles of flying - learning in just three hours what evolution took millions of years to achieve, according to research by Swedish scientists published on Wednesday.
Krister Wolff and Peter Nordin of Chalmers University of Technology built a robot with wings and then gave it random instructions through a computer at the rate of 20 per second.
Each instruction produced a small movement - the robot's wings could move up and down, forwards and backwards, and twist in either direction, the research published in Britain's New Scientist magazine said.
The robot was attached to two vertical poles to enable it to move up and down, and the metre-long wings were made from balsa-wood covered with a light plastic film.
Continues Below ↓
The programme instructed the robot its aim was to produce maximum lift, but had no pre-programmed data on the concept of flapping or how to do it.
At first the robot produced only twitching and jerking movements but gradually it succeeded in getting off the ground.
Feedback from a movement detector told the programme how successful each combination of instructions tried had been, enabling it to evolve by ditching unsuccessful ones and pairing up new combinations of the ones that produced most lift.
Cheating was one strategy tried and rejected during the process of artificial evolution - at one point the robot simply stood on its wing tips and later it climbed up on some objects that had been accidentally left nearby.
But after three hours the robot discovered a flapping technique - rotating its wings through 90 degrees, raising them, then twisting back to the horizontal before pushing back down.
"This tells us that this kind of evolution is capable of coming up with flying motion," said Peter Bentley, an evolutionary computer expert at University College, London.
However, the robot could not actually fly because it was too heavy for its electrical motor.
"There's only so much that evolution can do," Bentley said.
|
StumbleUpon
Reddit
Digg
Delicious
Newsvine
Facebook
muti
Twitter
