This simulation demonstrates the evolution of creatures that are controlled by a "brain" which is a recurrent neural network. The brain gets as a input several values, such as a vector that pointing to the nearest food (green dot) and sense of sight that identify poison (gray dot), and produces an output, which is, for example, the velocity (both direction and magnitude) of the creature, color, etc. Sensors allows each creature to sense its environment, for example the color of the creatures around him.
The aim of the program is to see interesting behavior evolve, without an artificial "fitness function". each creature losses some energy when it moves or hits another creature, and can gain some energy by eating the food. As one can notice, after few hours of running behaviors like steering towards food and avoiding from obstacles (gray dots).
The motivation for this program came from the "scriptbots" project:
https://sites.google.com/site/scriptbotsevo/
