Robot imitates itself when 'wounded'

A team of international scientists has created robots with special software that allows them to find ways to adapt to injuries in just 2 minutes.

Scientists from the University of Pierre and Marie Curie (France) and the University of Wyoming (USA) say they have built this robot based on the mechanism of self-recovery of animals.

When an animal is injured in the leg, it can adapt by limping, moving the focus to the other leg or some other strategy.


Robot can still go on injured.

For example, a dog with only three legs can still catch beetles, or if someone sprains their ankles, they can still find a way to move on.

"When injured, animals have intuitive ways of dealing with situations," said co-author Jean-Baptiste Mouret.

"These intuitions allow them to choose a number of different ways to try, and after these tests they will pick the most advantageous." We created robots that could do the same, " he explained. .

Before the operation, the robot uses a new algorithm to create a detailed map of the space. This allows it to develop a certain intuition about the behaviors it can perform and the predominance of such behaviors.

Basically, the robot can build a library of different motions and set up body parts that it can rely on if injured, even if it is broken or missing.

Picture 1 of Robot imitates itself when 'wounded'
Scientists hope to help robots "survive" longer when doing dangerous tasks such as rescuing people in fires or earthquakes - (Photo: University of Pierre and Marie Curie).

However, this process will take too long and potentially damage the equipment, so scientists have already mapped it using computer simulations.

By doing so, they can test and map over 13,000 different paths for the robot, including when "the robot's leg is broken, broken and fallen, and the robot's arm is broken," the researchers said. assist.

Scientists call this process "an intelligent algorithm and test , " which ultimately allows the robot to adapt to situations in up to two minutes.

With this research, they hope to help robots "survive" longer when doing dangerous tasks such as saving people in fires or earthquakes.