Winged Robot Learns to Fly
Krister Wolff and Peter Nordin of Chalmers University of Technology in Gothenburg, Sweden, built a winged robot and set about testing whether it could learn to fly by itself, without any pre-programmed data on what flapping is or how to do it.
This tells us that this kind of evolution is capable of coming up with flying motion, says Peter Bentley, who works on evolutionary computing at University College London. But while the robot had worked out how best to produce lift, it was not about to take off. Theres only so much that evolution can do, Bentley says. This thing is never going to fly because the motors will never have the strength to do it, he says.
The team attached the robot to two vertical rods, so it could slide up and down. At the start of a test, the robot was suspended by an elastic band. A movement detector measured how much lift, if any, the robot produced for any given movement.
Feedback from the movement detector let the program work out which sets of instructions were best at producing lift. The most successful ones were paired up and offspring sets of instructions were generated by swapping instructions randomly between successful pairs. These next-generation instructions were then sent to the robot and evaluated before breeding a new generation, and the process was repeated.
A. The two professors of CUT built the winged robot.
C. The two professors of CUT programmed the data on how the robot flapped its wings.
2.How did the robot behave at the beginning of the test?
B. It twitched but gradually gained height.
D. It landed not long after the test.
A. The winged robot could never really fly.
C. The winged robot should go through further evolution before it could fly.
4.What measured how much lift the robot produced?
C. An elastic band. D. Both B and C.
A. Pairing up successful instructions.
C. Generating new sets of instructions for evaluation.
Krister Wolff and Peter Nordin of Chalmers University of Technology in Gothenburg, Sweden, built a winged robot and set about testing whether it could learn to fly by itself, without any pre-programmed data on what flapping is or how to do it.
This tells us that this kind of evolution is capable of coming up with flying motion, says Peter Bentley, who works on evolutionary computing at University College London. But while the robot had worked out how best to produce lift, it was not about to take off. Theres only so much that evolution can do, Bentley says. This thing is never going to fly because the motors will never have the strength to do it, he says.
The team attached the robot to two vertical rods, so it could slide up and down. At the start of a test, the robot was suspended by an elastic band. A movement detector measured how much lift, if any, the robot produced for any given movement.
Feedback from the movement detector let the program work out which sets of instructions were best at producing lift. The most successful ones were paired up and offspring sets of instructions were generated by swapping instructions randomly between successful pairs. These next-generation instructions were then sent to the robot and evaluated before breeding a new generation, and the process was repeated.
A. The two professors of CUT built the winged robot.
C. The two professors of CUT programmed the data on how the robot flapped its wings.
2.How did the robot behave at the beginning of the test?
B. It twitched but gradually gained height.
D. It landed not long after the test.
A. The winged robot could never really fly.
C. The winged robot should go through further evolution before it could fly.
4.What measured how much lift the robot produced?
C. An elastic band. D. Both B and C.
A. Pairing up successful instructions.
C. Generating new sets of instructions for evaluation.
收藏