Evolving Neural Networks Of Bipedal Creatures
Bipedal creatures evolve to run to the right as fast as possible. Creatures that move to the right the fastest will have the higher fitness. Each bipedal creatures has 2 limbs consiting of a thigh and a leg. Each thigh is connected to the body with a joint. And each leg is connected to the thigh with a joint. Thus, the total number of joints are 4. Each joint is limited to 90 degrees of motion (45 left, 45 right). Creature takes damage if its limbs are not moving, and is killed if the body or the thighs touch the ground.
The four main classes that represent the contents of the research paper.
NEAConsultor.cs: Keeps track of new genes discovered in the world, and has a master genome of all discovered genes.
http://pastebin.com/0EQbnAwL
NEATNet.cs: Takes care of genome storage, genome mutation, crossover reproduction, phenotype activation. etc.
http://pastebin.com/KqW0SrK3
NEATGene.cs: The gene....obviously lol. Has in node, out node, weighted connection.
http://pastebin.com/h3AUE9Nj
NEATNode.cs: The node defined by node value and it's type.
http://pastebin.com/WABH56dT
Algorithm: NEAT (Neuro-evolution of augmenting Topology)
Number Of Inputs: Total 10 (1 Body angle, 4 joint angle, and 4 touch detection sensors, 1 bias)
Number Of Outputs: 4 (rotational speed)
Neural Network Storage Type: Gene List system as described by NEAT
Activation Function: Hyperbolic tangent approximation
Ranking Function: Explicit fitness sharing for each species
NEAT Paper by Kenneth O. Stanley and Risto Miikkulainen: http://nn.cs.utexas.edu/downloads/pap...