Update app.py

app.py

@@ -182,36 +182,53 @@ print("Q-Learning model trained!")
 
 def q_learning_path(q_values, start, goal, max_steps=100):
     """Find path using trained Q-Learning model"""
-    current_state = list(start)  # [x, y] format from frontend
-    path = [tuple(current_state)]
+    # Start: (x, y) tuple format
+    x, y = start
+    path = [start]
+    visited = set([start])
     
-    for _ in range(max_steps):
-        # Convert to [y, x] for q_values indexing
-        y, x = current_state[1], current_state[0]
+    for step in range(max_steps):
+        if (x, y) == goal:
+            return path
         
-        if tuple(current_state) == goal:
-            break
-        
-        # Get best action
+        # Get best action from Q-table [y, x, action]
         action = np.argmax(q_values[y, x])
         
-        # Apply action
+        # Apply action (0=up, 1=right, 2=down, 3=left)
+        new_x, new_y = x, y
         if action == 0 and y > 0:  # up
-            current_state[1] -= 1
+            new_y = y - 1
         elif action == 1 and x < COLS - 1:  # right
-            current_state[0] += 1
+            new_x = x + 1
         elif action == 2 and y < ROWS - 1:  # down
-            current_state[1] += 1
+            new_y = y + 1
         elif action == 3 and x > 0:  # left
-            current_state[0] -= 1
+            new_x = x - 1
         
-        path.append(tuple(current_state))
+        # Check if we're making progress
+        if (new_x, new_y) in visited:
+            # If stuck in loop, try alternative action
+            q_vals = q_values[y, x].copy()
+            q_vals[action] = -np.inf  # Don't use same action
+            action = np.argmax(q_vals)
+            
+            # Try alternative
+            new_x, new_y = x, y
+            if action == 0 and y > 0:
+                new_y = y - 1
+            elif action == 1 and x < COLS - 1:
+                new_x = x + 1
+            elif action == 2 and y < ROWS - 1:
+                new_y = y + 1
+            elif action == 3 and x > 0:
+                new_x = x - 1
         
-        # Prevent infinite loops
-        if len(path) > max_steps:
-            break
+        x, y = new_x, new_y
+        path.append((x, y))
+        visited.add((x, y))
     
-    return path if path[-1] == goal else None
+    # Return path even if goal not reached exactly
+    return path if len(path) > 1 else None
 
 def visualize_path(start_x, start_y, goal_x, goal_y, algorithm):
     start = (int(start_x), int(start_y))