src/main.py

@@ -63,12 +63,6 @@ def main():
 
     running = True
 
-    # ruchy agenta:
-    # 0 - up
-    # 1 - right
-    # 2 - down
-    # 3 - left
-
     while running:
         for event in pg.fastevent.get():
             if event.type == pg.QUIT:
@@ -76,9 +70,9 @@ def main():
             elif event.type == pg.KEYDOWN:
                 if event.key == pg.K_t:
                     print('Starting to clear the sector')
-                    while env.mine_count:
+                    # while env.mine_count:
                     print('-' * 20)
-                        # path, actions = breadth_first_search(env.field, agent.x, agent.y, agent.direction)
+
                     goal = breadth_first_search(env.field, agent.x, agent.y, agent.direction, True)
                     path, actions = a_star(env.field, agent.x, agent.y, agent.direction, goal)
 

src/search_algoritms/BFS.py

@@ -2,8 +2,10 @@ import queue
 from typing import List
 
 from src.tile import Tile
+from const import model, IMAGES
 from search_algoritms.node import Node
 from .helpers import successor, get_path_actions
+from machine_learning.neural_network.learning import prediction
 
 
 def breadth_first_search(
@@ -20,7 +22,8 @@ def breadth_first_search(
     while not node_queue.empty():
         node = node_queue.get()
 
-        if field[node.y][node.x].mine:
+        img_path = IMAGES[field[node.y][node.x].number].path
+        if field[node.y][node.x].occupied and prediction(img_path, model) == 'mine':
             return get_path_actions(node) if not a_star else node.x, node.y
 
         explored.append(node)

src/search_algoritms/helpers.py

@@ -3,8 +3,9 @@ from typing import List, Tuple
 import pygame as pg
 
 from src.tile import Tile
-from src.const import ROCK_INDEXES
+from src.const import IMAGES, model
 from src.search_algoritms.node import Node
+from machine_learning.neural_network.learning import prediction
 
 
 def get_path_actions(node: Node):
@@ -31,9 +32,18 @@ def successor(field: List[List[Tile]], x: int, y: int, direction: int):
     neighbours.append((x, y, (direction - 1) % 4, pg.K_a))
     neighbours.append((x, y, (direction + 1) % 4, pg.K_d))
 
-    if coord == 'x' and 0 <= x + shift <= 9 and field[y][x + shift].number not in ROCK_INDEXES:
+    if coord == 'x' and 0 <= x + shift <= 9:
+        img_path = IMAGES[field[y][x + shift].number].path
+        impassable = True if field[y][x + shift].occupied and prediction(img_path, model) == 'rock' else False
+
+        if not impassable:
             neighbours.append((x + shift, y, direction, pg.K_w))
-    elif coord == 'y' and 0 <= y + shift <= 9 and field[y + shift][x].number not in ROCK_INDEXES:
+
+    elif coord == 'y' and 0 <= y + shift <= 9:
+        img_path = IMAGES[field[y + shift][x].number].path
+        impassable = True if field[y + shift][x].occupied and prediction(img_path, model) == 'rock' else False
+
+        if not impassable:
             neighbours.append((x, y + shift, direction, pg.K_w))
 
     return neighbours