-
An early generation struggling to get past the first obstacle.
-
A later-generation bird with the favorable mutations of many past generation birds helping him to successfully navigate the obstacles.
Inspiration
We were inspired by and attracted to the idea of neuroevolution because of the biological similarities to actual thinking that we noticed.
What it does
The program spawns a lot of bird sprites that are nearly identical but for slight "mutations" in their patterns of movement, that are passed down from more successful birds to the next generation of birds.
How we built it
We just used python to program a "fitness" value that increased over time as a bird flew, then used the fitness value in a weighting system to determine how the next generation would react (jump, or don't) to inputs such as the horizontal and vertical distances to obstacles.
Challenges we ran into
We got an error that took us a long time to fix with some syntax.
Accomplishments that we're proud of
Actually finishing a working product.
What we learned
A lot of things about AI.
What's next for Neuroevolution for Flappy Bird
Hopefully winning something.
Log in or sign up for Devpost to join the conversation.