acm-header
Sign In

Communications of the ACM

ACM TechNews

Robotic Elephant Trunk Can Learn Tasks on Its Own


View as: Print Mobile App Share: Send by email Share on reddit Share on StumbleUpon Share on Hacker News Share on Tweeter Share on Facebook
The robotic elephant trunk.

Researchers at the University of Tubingen in Germany used three-dimensional printing to create a robotic elephant trunk that utilizes artificial intelligence to emulate how sensory input triggers synaptic chain reactions in organic brains.

Credit: New Scientist

Researchers at Germany's University of Tubingen used three-dimensional (3D) printing to create a robotic elephant trunk that utilizes artificial intelligence (AI) to emulate how sensory input triggers synaptic chain reactions in organic brains.

Tubingen's Sebastian Otte and colleagues assembled the device from modules with gear-driving motors that tilt up to 40 degrees in two axes.

The AI was trained on examples of the motor inputs required to move the trunk in certain ways, and tests showed it could direct the tip of the trunk within less than a centimeter from a target.

The robot is a proof of concept of a spiking neural network algorithm, which works like an actual brain in which certain inputs cause a chain reaction of firing synapses, while requiring orders of magnitude less computational power and energy.

From New Scientist
View Full Article

 

Abstracts Copyright © 2021 SmithBucklin, Washington, DC, USA


 

No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account