M. Oubbati, P. Levi, and M. Schanz (Germany)
Mobile robot, Recurrent Neural Networks, Meta-learning, and Adaptive control
In this paper, we demonstrate the ability of a single fixed weight RNN to act as a dynamic controller for several (here 3) distinct wheeled mobile robots, without exact knowledge about their dynamics parameters. The controller is properly trained to exhibit adaptive behaviour after its weights have been fixed. This capability is a natural consequence of prior `meta-learning' used recently in the area of RNN.
Important Links:
Go Back
