Using learning algorithms individual swimmers form schools by overcoming hydordynamics interactions

What is it about?

Swimmers disturb the fluid around them, and as such, nearby swimmers must compensate for these disturbances to perform coordinated swimming. We show that reinforcement learning algorithms (a form of unsupervised learning) can enable such a task for the first time.

Featured Image

Photo by Jonny Clow on Unsplash

Photo by Jonny Clow on Unsplash

Why is it important?

This demonstrates that energy deposited in the wakes of fish and other swimming organisms can be harnessed adaptively in order to achieve certain goals.