Intelligent Systems
Note: This research group has relocated.

Shaping in Practice: Training Wheels to Learn Fast Hopping Directly in Hardware

3 February 2018

01:59

We show a simple case of using "training wheels", temporary hardware modifications, to shape the reward landscape and make learning easier. Full paper to appear in ICRA 2018, available on arxiv at: https://arxiv.org/abs/1709.10273

=