Learning to Perform a Perched Landing Using Deep Reinforcement Learning (link.springer.com) 20 points by cjCamel 8y ago ↗ HN
[–] cjCamel 8y ago ↗ Associated video here: https://www.youtube.com/watch?v=uU0WDWV6au8&feature=youtu.be
[–] mendeza 8y ago ↗ Thats amazing! I wonder how they made the training data to teach the drone how to land, I can assume it is challenging to fly a drone hundreds of times to get a good landing. Unless they had a different model that they could sample from. [–] uoaei 8y ago ↗ It is reinforcement learning, i.e., necessarily unsupervised.
3 comments
[ 329 ms ] story [ 122 ms ] thread