University of Wyoming and Pierre and Marie Curie University researchers are developing robots that can become more self-sufficient by finding ways to adapt to their environment and keep moving after an injury.
For example, a robot sent on a search and rescue mission after an earthquake would be able to handle unexpected damage to its body. The robot learns with an evolutionary algorithm, which means that it will run repeated trial and error steps to find a new way to get around.
As part of the study, the robots did not understand what was wrong with them, and the researchers did not try to anticipate anything about the damage they would sustain. Although many robotic systems have sensors that could identify problems, they cannot correct the problems, so the researchers hypothesize that a robot is better served finding a new way to move around through experimentation rather than sensor data.
"Almost all animals are built to adapt to a small injury," says former Pierre and Marie Curie University researcher Jean-Baptiste Mouret.
The researchers detail a six-legged robot that, after sustaining an injury to one limb, starts trying new ways of moving, and in about 40 seconds regains 96 percent of its speed.
From Technology Review
View Full Article
Abstracts Copyright © 2015 Information Inc., Bethesda, Maryland, USA