Genetic Alghoritm implemented, Tree moved

collect

@@ -24,11 +24,11 @@ edge [fontname="helvetica"] ;
 6 -> 10 ;
 11 [label="garbage_weight <= 0.612\ngini = 0.094\nsamples = 61\nvalue = [3, 58]\nclass = no-collect"] ;
 10 -> 11 ;
-12 [label="garbage_type <= 2.0\ngini = 0.5\nsamples = 2\nvalue = [1, 1]\nclass = collect"] ;
+12 [label="space_occupied <= 0.382\ngini = 0.5\nsamples = 2\nvalue = [1, 1]\nclass = collect"] ;
 11 -> 12 ;
-13 [label="gini = 0.0\nsamples = 1\nvalue = [1, 0]\nclass = collect"] ;
+13 [label="gini = 0.0\nsamples = 1\nvalue = [0, 1]\nclass = no-collect"] ;
 12 -> 13 ;
-14 [label="gini = 0.0\nsamples = 1\nvalue = [0, 1]\nclass = no-collect"] ;
+14 [label="gini = 0.0\nsamples = 1\nvalue = [1, 0]\nclass = collect"] ;
 12 -> 14 ;
 15 [label="garbage_type <= 2.5\ngini = 0.065\nsamples = 59\nvalue = [2, 57]\nclass = no-collect"] ;
 11 -> 15 ;
@@ -50,11 +50,11 @@ edge [fontname="helvetica"] ;
 5 -> 23 ;
 24 [label="gini = 0.0\nsamples = 2\nvalue = [0, 2]\nclass = no-collect"] ;
 23 -> 24 ;
-25 [label="odour_intensity <= 8.841\ngini = 0.219\nsamples = 8\nvalue = [7, 1]\nclass = collect"] ;
+25 [label="days_since_last_collection <= 22.0\ngini = 0.219\nsamples = 8\nvalue = [7, 1]\nclass = collect"] ;
 23 -> 25 ;
 26 [label="gini = 0.0\nsamples = 6\nvalue = [6, 0]\nclass = collect"] ;
 25 -> 26 ;
-27 [label="days_since_last_collection <= 22.0\ngini = 0.5\nsamples = 2\nvalue = [1, 1]\nclass = collect"] ;
+27 [label="odour_intensity <= 8.841\ngini = 0.5\nsamples = 2\nvalue = [1, 1]\nclass = collect"] ;
 25 -> 27 ;
 28 [label="gini = 0.0\nsamples = 1\nvalue = [1, 0]\nclass = collect"] ;
 27 -> 28 ;

garbage_truck.py

@@ -54,6 +54,8 @@ class GarbageTruck:
     def next_destination(self):
         
         for i in range(len(self.request_list)):
+            if(self.request_list==[]):
+                break
             request = self.request_list[i]
 
             #nie ma miejsca w zbiorniku lub za ciężkie śmieci
@@ -64,25 +66,10 @@ class GarbageTruck:
             if heuristicfn(request.x_pos, request.y_pos, self.dump_x, self.dump_y) // 50 * 200 > self.fuel:
                 continue
             
-
+            self.request_list.pop(i)
-
+            self.free_space -= request.volume
-            distance = heuristicfn(self.rect.x, self.rect.y, request.x_pos, request.y_pos) // 50
+            self.weight_capacity -= request.weight
-
+            return request.x_pos, request.y_pos   
-            r = [
-                self.fuel, 
-                distance,
-                request.volume,
-                request.last_collection,
-                request.is_paid,
-                request.odour_intensity,
-                request.weight,
-                request.type
-            ]
-            if self.clf.predict([r]) == True:
-                self.request_list.pop(i)
-                self.free_space -= request.volume
-                self.weight_capacity -= request.weight
-                return request.x_pos, request.y_pos   
         return self.dump_x, self.dump_y
             
 
@@ -96,7 +83,10 @@ class GarbageTruck:
             self.weight_capacity_mixed = MAX_WEIGHT_MIXED
             self.weight_capacity_glass = MAX_WEIGHT_GLASS
             self.weight_capacity_paper = MAX_WEIGHT_PAPER
-        request = self.request_list[0]
+        if self.request_list==[]:
+            return 1
+        else:
+             request = self.request_list[0]
         if garbage_type == "glass":
             if request.weight > self.weight_capacity_glass:
                 return 1

genetic.py

@@ -0,0 +1,162 @@
+import pygame
+from treelearn import treelearn
+import loadmodel
+from astar import astar
+from state import State
+import time
+from garbage_truck import GarbageTruck
+from heuristicfn import heuristicfn
+from map import randomize_map
+from heuristicfn import heuristicfn
+import pygame as pg
+import random
+from request import Request
+
+def determine_fitness(requests_list):
+    distances = []
+    for i in range(len(requests_list)+1): #from: request_list[i].x_pos and .y_pos
+        temp = []
+        for j in range(len(requests_list)+1):
+            if j<i:
+                temp.append('-')
+            elif j==i:
+                temp.append(0)
+            elif j>i:
+                if i==0:
+                    dist = heuristicfn(0, 0, requests_list[j-1].x_pos, requests_list[j-1].y_pos)
+                    temp.append(dist)
+                else:
+                    dist = heuristicfn(requests_list[i-1].x_pos, requests_list[i-1].y_pos, requests_list[j-1].x_pos, requests_list[j-1].y_pos)
+                    temp.append(dist)
+        distances.append(temp)
+    return(distances)
+
+
+def perform_permutation(obj_list, perm_list):
+    result = [None] * len(obj_list)
+
+    for i, index in enumerate(perm_list):
+        result[int(index)-1] = obj_list[i-1]
+    return result
+
+def apply_genetic(request_list):
+    print("Genetic algorithm started")
+
+    distances = determine_fitness(request_list)
+    population_size = 12
+    num_generations = 8
+    mutation_rate = 0.3
+    NUM = len(distances)
+
+    def initialize_population():
+        population = []
+        for _ in range(population_size):
+            chromosome = ['0']
+            while True:
+                if len(chromosome) == NUM:
+                    chromosome.append('0')
+                    break
+
+                temp = random.randint(1, NUM-1)
+                temp_str = str(temp)
+                if temp_str not in chromosome:
+                    chromosome.append(temp_str)
+            population.append(chromosome)
+        return population
+    
+    def calculate_route_length(route):
+        length = 0
+        for i in range(len(route)-1):
+            p = int(route[i])
+            q = int(route[i + 1])
+            length += distances[int(min(p,q))][int(max(p,q))]
+        return length
+
+    def calculate_fitness(population):
+        fitness_scores = []
+        for chromosome in population:
+            fitness_scores.append(1 / calculate_route_length(chromosome))
+        return fitness_scores
+
+    def parents_selection(population, fitness_scores):
+        selected_parents = []
+        for _ in range(len(population)):
+            candidates = random.sample(range(len(population)), 2)
+            fitness1 = fitness_scores[candidates[0]]
+            fitness2 = fitness_scores[candidates[1]]
+            selected_parent = population[candidates[0]] if fitness1 > fitness2 else population[candidates[1]]
+            selected_parents.append(selected_parent)
+        return selected_parents
+
+
+
+    def david_crossover(parent1, parent2):
+        start_index = random.randint(1, len(parent1)-3)
+        end_index = random.randint(start_index+1, len(parent1)-2)
+        parent1_chain = parent1[start_index:end_index+1]
+        parent2_letters = []
+        for trash in parent2[1:-1]:
+            if trash not in parent1_chain:
+                parent2_letters.append(trash)
+        child = [parent2[0]]+parent2_letters[0:start_index] + parent1_chain + parent2_letters[start_index:]+[parent2[-1]]
+        """         print('PARENTS: ')
+        print(parent1)
+        print(parent2)
+        print('CHILDS:')
+        print(child) """
+        return child
+
+    def mutation(chromosome):
+        index1 = random.randint(1, len(chromosome)-2)
+        index2 = random.randint(1, len(chromosome)-2)
+        chromosome[index1], chromosome[index2] = chromosome[index2], chromosome[index1]
+        return chromosome
+
+    def genetic_algorithm():
+        population = initialize_population()
+        for _ in range(num_generations):
+            fitness_scores = calculate_fitness(population)
+            parents = parents_selection(population, fitness_scores)
+            offspring = []
+            for i in range(0, len(parents), 2):
+                parent1 = parents[i]
+                parent2 = parents[i+1]
+                child1 = david_crossover(parent1, parent2)
+                child2 = david_crossover(parent2, parent1)
+                offspring.extend([child1, child2])
+            population = offspring
+            for i in range(len(population)):
+                if random.random() < mutation_rate:
+                    population[i] = mutation(population[i])
+        return population
+
+
+    best_route = None
+    best_length = float('inf')
+    population = genetic_algorithm()
+    for chromosome in population:
+        length = calculate_route_length(chromosome)
+        if length < best_length:
+            best_length = length
+            best_route = chromosome
+
+    print("Permutation chosen: ", best_route)
+    print("Its length:", best_length)
+    permuted_list = perform_permutation(request_list, best_route[1:-1])
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+

main.py

@@ -7,6 +7,8 @@ import time
 from garbage_truck import GarbageTruck
 from heuristicfn import heuristicfn
 from map import randomize_map
+from tree import apply_tree
+from genetic import apply_genetic
 
 
 pygame.init()
@@ -38,6 +40,8 @@ def main():
     clock = pygame.time.Clock()
     run = True
     fields, priority_array, request_list, imgpath_array = randomize_map()
+    apply_tree(request_list)
+    apply_genetic(request_list)
     agent = GarbageTruck(0, 0, pygame.Rect(0, 0, 50, 50), 0, request_list, clf)   # tworzenie pola dla agenta
     low_space = 0
     while run:

tree.py

@@ -0,0 +1,35 @@
+import pygame
+from treelearn import treelearn
+import loadmodel
+from astar import astar
+from state import State
+import time
+from garbage_truck import GarbageTruck
+from heuristicfn import heuristicfn
+from map import randomize_map
+from heuristicfn import heuristicfn
+import pygame as pg
+import random
+from request import Request
+
+def apply_tree(request_list):
+
+    print("Przed zastosowaniem drzewa na liście jest śmieci: ", len(request_list))
+
+    for address in request_list:
+        r = [
+                    0, 
+                    0,
+                    address.volume,
+                    address.last_collection,
+                    address.is_paid,
+                    address.odour_intensity,
+                    address.weight,
+                    address.type
+                ]
+        clf = treelearn()
+        if clf.predict([r]) == False:
+            request_list.pop(request_list.index(address))
+
+    print("Po zastosowaniu drzewa na liście jest śmieci: ", len(request_list))
+    return request_list