Conversation

Some unexpectedly beautiful images from a 1998 paper on a machine learning bicycle learning to drive.
Here, the paths it took as it learned to balance (left) and drive toward a target slightly off to the right.
https://www.semanticscholar.org/paper/Learning-to-Drive-a-Bicycle-Using-Reinforcement-and-Randl%C3%B8v-Alstr%C3%B8m/10bad197f1c1115005a56973b8326e5f7fc1031c