// Update is called once per frame
void Update()
{
//Used to track the best performing spawn
double bestFitness = 0;
int bestSpawn = 0;
//TimeScale
if (Input.GetKeyDown(KeyCode.Comma))
{
Time.timeScale *= .5f;
}
if (Input.GetKeyDown(KeyCode.Period))
{
Time.timeScale *= 2;
}
int species = 0;
int genome = 0;
//Update each spawn in the population, used to increase their fitness as needed and have them respond to the inputs
for (int i = 0; i < population; i++)
{
GameObject closestFood = foods[0];
float distance = Vector3.Distance(closestFood.transform.position, spawns[i].transform.position); //distance to closest food
SpriteRenderer sprite = spawns[i].GetComponent <SpriteRenderer>(); //get spawn sprite
sprite.color = genAlg.species[species].color;
for (int j = 0; j < numFood; j++) //Finds the closest food item
{
float newDistance = Vector3.Distance(spawns[i].transform.position, foods[j].transform.position);
if (newDistance < distance)
{
distance = newDistance;
closestFood = foods[j];
}
}
//relative position of spawn to closest food
Vector2 relPos = spawns[i].transform.position - closestFood.transform.position;
relPos.Normalize();
Vector2 relRotation = spawns[i].transform.up;
relRotation.Normalize();
//---------------Create the input list--------------------------------------------\\
List <double> input = new List <double> {
relPos.x, relPos.y, relRotation.x, relRotation.y, 1
};
//---------------Get the ouput list by inputing the input list--------------------\\
List <double> output = genAlg.species[species].population[genome].Run(input);
//----------------Applying the output to te spawn---------------------------------\\
spawns[i].transform.Rotate(new Vector3(0, 0, (float)(output[0] - output[1]) * Time.deltaTime * 100));
spawns[i].transform.Translate(Vector2.up * speedAdj * Time.deltaTime);
//---------------Increasing Fitness-------------------------------------------------\\
if (distance < 1f)
{
genAlg.species[species].population[genome].fitness += genAlg.species[species].population[genome].fitness * .1F + 1;
//move food particle
closestFood.transform.position = new Vector3(Random.Range(0, gameArea.x), Random.Range(0, gameArea.y));
}
//Some informational text, and allowing for the search of the most fit spawn
if (i == 0)
{
infoText.text = "Generation: " + genAlg.generation + "\nInputs: relPosX, relPosY, relRotaionX, relRotationY\nOutputs: rightRotation, leftRotation"
+ "\nTime: " + Time.time.ToString("0")
+ "\nTime Scale: " + Time.timeScale
+ "\nSpecies Count: " + genAlg.species.Count;
bestFitness = genAlg.species[species].population[genome].fitness;
bestSpawn = i;
//----------------Use the first spawn as the selected object in visual representation--------------//
if (updateGraph)
{
graph.NodeUpdate(genAlg.species[randomWatchSpecies].population[randomWatchGenome]);
}
}
else
{
if (bestFitness < genAlg.species[species].population[genome].fitness)
{
bestFitness = genAlg.species[species].population[genome].fitness;
sprite = spawns[bestSpawn].GetComponent <SpriteRenderer>();
sprite.color = genAlg.species[species].color;
bestSpawn = i;
sprite = spawns[i].GetComponent <SpriteRenderer>();
sprite.color = Color.yellow;
}
}
if (species == randomWatchSpecies && genome == randomWatchGenome)
{
randomWatchSpawn = i;
}
genome++;
if (genome >= genAlg.species[species].population.Count)
{
species++;
genome = 0;
}
//Border control
if (spawns[i].transform.position.y < 0)
{
spawns[i].transform.position = new Vector3(spawns[i].transform.position.x, gameArea.y);
}
else if (spawns[i].transform.position.y > gameArea.y)
{
spawns[i].transform.position = new Vector3(spawns[i].transform.position.x, 0);
}
if (spawns[i].transform.position.x < 0)
{
spawns[i].transform.position = new Vector3(gameArea.x, spawns[i].transform.position.y);
}
else if (spawns[i].transform.position.x > gameArea.x)
{
spawns[i].transform.position = new Vector3(0, spawns[i].transform.position.y);
}
}
//More Stats
genAlg.CalculateBestWorstAvTot();
infoText.text += "\nBest Fitness: " + genAlg.bestFitness + "\nAverage Fitness: " + genAlg.averageFitness + "\nLowest Fitness: " + genAlg.worstFitness;
fitnessGraph.UpdateFitnessGraph(genAlg.generation, genAlg.bestFitness, genAlg.averageFitness);
//Change first spawn's color so it is recognizable
spawns[randomWatchSpawn].GetComponent <SpriteRenderer>().color = Color.cyan;
//----------------Creates the New Population using the fitness levels of the old one----------------//
if (Time.time > updateTime)
{
updateTime += populationResetTime;
//--------------The main function that creates the new population into the genAlg class--------//
genAlg.Epoch();
for (int i = 0; i < population; i++)
{
//change all of the spawns positions
spawns[i].transform.position = new Vector3(Random.Range(0, gameArea.x), Random.Range(0, gameArea.y));
//Update the spawns neural networks with the new weights
//nNetwork[i].PutWeights(genAlg.population[i].weights);
}
//change the location of all the food
for (int i = 0; i < numFood; i++)
{
foods[i].transform.position = new Vector3(Random.Range(0, gameArea.x), Random.Range(0, gameArea.y));
}
randomWatchSpecies = Random.Range(0, genAlg.species.Count);
randomWatchGenome = Random.Range(0, genAlg.species[randomWatchSpecies].population.Count);
graph.CreateGraph(genAlg.species[randomWatchSpecies].population[randomWatchGenome]);
fitnessGraph.UpdateSpeciesPopulationGraph(genAlg.species, genAlg.generation);
}
}
