A collection of Environments to Create, Train and Test different kinds of Artificial Intelligente algorithms to be used inside the Fast and Curious game, which you can find .

Inside some Missions of the game, there is a Car trying to run away from a Space-Time entity which wants to destroy it. The road is filled with various types of obstacles (which cause harm on impact) and collectables (the more collected the better).

Make the Car avoid ALL obstacles while collecting as many collectables as possible.

• Model: Neural Networks;
• Training Algorithm: ;
• Inputs: x and y position of the car, x and y distance to the 2 closest obstacles, x and y distance to the closest collectable;
• Results: Goal is easly achieved in under 5 minutes in terms of avoiding obstacles, but it could be improved in terms of collecting items. The algorithm was ran just for a maximum of 20 minutes, which means a better result could be achieved for longer training times.