# DoodleJumpAI

**Repository Path**: bit2atom/DoodleJumpAI

## Basic Information

- **Project Name**: DoodleJumpAI
- **Description**: A Neural Network that learns to play Doodle Jump trained using a Genetic Algorithm
- **Primary Language**: Unknown
- **License**: MIT
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2020-08-25
- **Last Updated**: 2020-12-19

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# DoodleJumpAI

## Download and Run
To run the program you will need [Processing](https://processing.org/)

## Doodle
### Neural Network
Each doodle contains a neural network. The neural network has an input layer of 5 neurons, 2 hidden layers of 4 neurons, and one output layer of 2 neurons. 
### Genetic Algorithm
A population of 200 doodles is created, each doodle with its own neural network. After all the doodles die some of the highest scoring ones are chosen to reproduce and create a new population of 200 doodles. This process is repeated for every generation.
### Vision
The doodle can see in 5 directions. In each of these directions the doodle can see the distance to a pad, if there is one in that direction.

![doodlevision](https://user-images.githubusercontent.com/36581610/48689204-5c8af600-eb97-11e8-8deb-e0391667e4d7.PNG)

![doodlejump-1](https://user-images.githubusercontent.com/36581610/50392983-e7c05400-0720-11e9-8c97-523f1e3687b7.gif)

![doodlejump-2](https://user-images.githubusercontent.com/36581610/50393244-c496a400-0722-11e9-9ee5-375d67511df3.gif)