guezoloic/SNAKE
1
0
Fork 0
mirror of https://github.com/Cpt-Adok/SNAKE.git synced 2026-01-25 10:34:06 +00:00
Code Issues Packages Projects Releases Wiki Activity

correction de probleme

Browse Source
This commit is contained in:
guezoloic
2024-05-31 10:54:08 +02:00
parent b053f78fc2
commit b40750c4ed
4 changed files with 76 additions and 69 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
src/IA/QLearning.java
View File

@@ -95,6 +95,7 @@ public class QLearning {
bestMouvement = mouvement;
}
}
return bestMouvement;
}
}

61
src/Main.java
View File

@@ -4,43 +4,50 @@ import IA.QTable;
import environnement.*;
import game.Terminal;
import personnage.*;
import tests.IATest;
public class Main {
public static void main(String[] args) {
Personnage.n = 4;
// Personnage.n = 4;
Map map = new Map(12, 22);
// Map map = new Map(12, 22);
// lancer en local
if (args.length < 2) {
Grid[][] grid = map.getGrid();
// // lancer en local
// if (args.length < 2) {
// Grid[][] grid = map.getGrid();
// QTable qTable = new QTable();
// qTable.getValues("res" + File.separator + "save" + File.separator + "learn.ser");
// QTable qTable1 = new QTable();
// qTable1.getValues("path_to_save_qtable1.ser");
// Avant de jouer contre l'ia, vous pouvez essayer de l'entrainer avec la fonction tests.IATest.learnIAvsIA()
// il jouera avec lui meme et mettra les sauvegardes dans le dossier learn.ser,
// QTable qTable2 = new QTable();
// qTable2.getValues("path_to_save_qtable2.ser");
// Attention lors de l'apprentissage, ne pas couper le processus sinon vous allez perdre toute vos donnees
Personnage[] personnages = new Personnage[] {
new Player(new int[] {2, 2}, "Philippe Etchebest"),
new Player(new int[] {grid[0].length - 3, grid.length - 3}, "Luke Skywalker"),
// new Robot("Robot", new int[] {grid[0].length - 3, grid.length - 3}),
// new IAQLearning(new int[] {grid[0].length - 3, grid.length - 3),
};
// // Avant de jouer contre l'ia, vous pouvez essayer de l'entrainer avec la fonction tests.IATest.learnIAvsIA()
// // il jouera avec lui meme et mettra les sauvegardes dans le dossier learn.ser,
// map.addObjectsRandomize(new Item[] {Item.FRAISE, Item.WALL}, 2);
// map.addObjects(Item.FRAISE, 2, 2);
// // Attention lors de l'apprentissage, ne pas couper le processus sinon vous allez perdre toute vos donnees
// Personnage[] personnages = new Personnage[] {
// new IAQLearning(new int[] {2, 2}, qTable1),
// // new Player(new int[] {2, 2}, "Philippe Etchebest"),
// // new Player(new int[] {grid[0].length - 3, grid.length - 3}, "Luke Skywalker"),
// // new Robot("Robot", new int[] {grid[0].length - 3, grid.length - 3}),
// new IAQLearning(new int[] {grid[0].length - 3, grid.length - 3}, qTable2),
// };
new Terminal(map, personnages).run();
}
// lancer en ligne
else {
Personnage[] personnages = new Personnage[] {
new Player(new int[] {0, 0}, "Philippe Etchebest"),
};
// // map.addObjectsRandomize(new Item[] {Item.FRAISE, Item.WALL}, 2);
// // map.addObjects(Item.FRAISE, 2, 2);
new Terminal(map, personnages).run(args[0], args[1]);
}
// new Terminal(map, personnages).run();
// }
// // lancer en ligne
// else {
// Personnage[] personnages = new Personnage[] {
// new Player(new int[] {0, 0}, "Philippe Etchebest"),
// };
// new Terminal(map, personnages).run(args[0], args[1]);
// }
IATest.learnIAvsIA();
}
}

83
src/tests/IATest.java
View File

@@ -1,9 +1,11 @@
package tests;
import java.io.File;
import java.util.Arrays;
import IA.QTable;
import IA.State;
import display.Display;
import environnement.Grid;
import environnement.Map;
import personnage.IAQLearning;
@@ -22,14 +24,14 @@ public class IATest {
double decay_rate = 0.995;
double minEpsilon = 0.01;
int totalEpisodes = 1000;
int totalEpisodes = 200;
Personnage.n = 2;
Personnage.n = 4;
for(int episode = 0; episode < totalEpisodes; episode++) {
QTable qTable = new QTable();
IAQLearning iaqLearning = new IAQLearning(new int[] {0, 0}, qTable, alpha, gamma, epsilon);
Map map = new Map(20, 20);
IAQLearning iaqLearning = new IAQLearning(new int[] {2, 2}, qTable, alpha, gamma, epsilon);
Map map = new Map(12, 22);
qTable.getValues(path);
@@ -72,37 +74,29 @@ public class IATest {
public static void learnIAvsIA() {
double alpha = 0.1;
double gamma = 0.9;
double epsilon = 0.1;
double[] epsilon = new double[] {1.0,};
double decay_rate = 0.995;
double minEpsilon = 0.01;
int totalEpisodes = 1000;
int maxEpisode = 1000;
Personnage.n = 4;
for (int episode = 0; episode < totalEpisodes; episode++) {
for (int episode = 0; episode < maxEpisode; episode++) {
QTable qTable = new QTable();
IAQLearning[] iaqLearnings = new IAQLearning[] {
new IAQLearning(new int[] {2, 2}, qTable, alpha, gamma, epsilon[0]),
new IAQLearning(new int[] {9, 19}, qTable, alpha, gamma, epsilon[1])
};
Map map = new Map(12, 22);
boolean isGameOver = false;
qTable.getValues(path);
while(true) {
for (int i = 0; i < iaqLearnings.length; i++) {
IAQLearning iaqLearning = iaqLearnings[i];
Map map = new Map(12, 22);
Grid[][] gridMap = map.getGrid();
Map mapIA = new Map(gridMap[0].length, gridMap.length);
mapIA.replaceGrid(gridMap);
IAQLearning[] iaqLearnings = new IAQLearning[] {
new IAQLearning(new int[] {2, 2}, qTable, alpha, gamma, epsilon),
new IAQLearning(new int[] {9, 19}, qTable, alpha, gamma, epsilon),
};
boolean isGameOver = false;
while(true) {
for (int personnages = 0; personnages < iaqLearnings.length; personnages++) {
IAQLearning iaqLearning = iaqLearnings[personnages];
Map mapIA = new Map(map.getGrid()[0].length, map.getGrid().length);
for (IAQLearning value : iaqLearnings) {
map.placePersonnages(value);
@@ -110,15 +104,25 @@ public class IATest {
State currentState = iaqLearning.getCurrentState(map.getGrid());
Mouvement mouvement = iaqLearning.bestMouvement(currentState);
iaqLearning.moveSnake(mouvement);
int[] coordinate = iaqLearning.getHeadCoordinate();
for (int[] snakeCoordinate : iaqLearnings[(i + 1) % 2].getCoordinate()) {
if (coordinate[0] == snakeCoordinate[0] && coordinate[1] == snakeCoordinate[1]) {
iaqLearning.receiveReward(currentState, mouvement, -10.0, currentState);
iaqLearnings[(i + 1) % 2].receiveReward(currentState, mouvement, 10.0, currentState);
if (map.isGameOver(coordinate) || iaqLearning.applyEffects(map.getEffect(coordinate))) {
iaqLearning.receiveReward(currentState, mouvement, -1000, currentState);
isGameOver = true;
break;
}
int value = (personnages + 1) % 2;
for (int[] snakeCoordinate : iaqLearnings[value].getCoordinate()) {
if (Arrays.equals(coordinate, snakeCoordinate)) {
iaqLearnings[value].receiveReward(currentState, mouvement, 1000, currentState);
iaqLearning.receiveReward(currentState, mouvement, -500, currentState);
isGameOver = true;
break;
}
}
@@ -126,24 +130,19 @@ public class IATest {
mapIA.placePersonnages(iaqLearning);
State nextState = iaqLearning.getCurrentState(mapIA.getGrid());
iaqLearning.receiveReward(currentState, mouvement, -0.1, nextState);
iaqLearning.receiveReward(currentState, mouvement, -0.01, nextState);
iaqLearning.increaseRound();
mapIA.clearMap();
map.clearMap();
}
if(isGameOver) break;
qTable.save(path);
System.out.println("Episode: " + episode + " States: " + qTable.getqValues().size());
}
qTable.save(path);
for (int i = 0; i < epsilon.length; i++) {
epsilon[i] = Math.max(minEpsilon, epsilon[i] * decay_rate);
}
System.out.println("Episode: " + episode + " | Robot 1 States: " + qTable.getqValues().size());
}
}
}