Communications of the ACM
Engineers Teach AI to Navigate Ocean with Minimal Energy
For the purposes of this project, the team wrote software that can be installed and run on a Teensya 2.4-by-0.7-inch microcontroller that anyone can buy for less than $30 on Amazon and only uses about a half-watt of power.
Credit: Peter Gunnarson
Artificial intelligence (AI) technology developed by researchers at the California Institute of Technology (Caltech), Harvard University, and Switzerland's ETH Zurich aims to enable autonomous drones exploring the deep ocean to navigate using currents instead of fighting them.
The researchers used reinforcement learning networks to train the AI take advantage of low-velocity regions in the wake of vortices moving in opposite directions to navigate to the target location using minimal power.
The simulated swimmer was given access only to information about water currents at its immediate location, but learned how to coast toward the target.
Said Caltech's John O. Dabiri, "We were initially just hoping the AI could compete with navigation strategies already found in real swimming animals, so we were surprised to see it learn even more effective methods by exploiting repeated trials on the computer."
From California Institute of Technology
View Full Article
Abstracts Copyright © 2021 SmithBucklin, Washington, DC, USA
No entries found