Researchers at the University of Lorraine have developed a “Damage Reflex” (also known as D-Reflex) system.
He is held by a humanoid robot TALOS who leans against a wall when one of his legs breaks, like a human who has just lost his balance.
The neural network-based system uses its experience (in this case, 882,000 training simulations) to quickly find a point on the wall most likely to provide stability.The robot does not need to know how it was damaged and can reach it as fast as a person.
The result, as IEEE Spectrum points out, is the anti-comedy you’d expect.
The D-Reflex applied in a robot
Instead of falling to the floor, the robot leans against the wall like someone who just sprained their ankle.
It’s not particularly elegant, requiring the robot to stop its hand the moment it makes contact, but it’s effective in three out of four tests.
D-Reflex is not guaranteed to prevent a fall, in part because it cannot account for all possible positions or surfaces.
Nor does it help the robot recover once it averts catastrophe: not seeing the automaton limping along a wall until help is found.
The current approach is also based on a stationary bot and won’t help if an actuator fails midway.
However, the researchers hope to create a system that is useful in motion and envision robots that can grab chairs and other complex objects when a fall is imminent.
This could save the cost of replacing worker bots that would otherwise plunge to their doom, and could lead to more “natural” bots learning to use their environments to their advantage.
Subscribe to our Unocero Youtube channel to learn more about technology.